an open-source alternative to the dots, bots and muses of the world, run an office of claude code/codex like agents on your laptop, sandboxes or anywhere, uses your existing subscriptions
HTML agent-harnessagent-orchestrationagentsai-agentsautonomous-agentsclaude-codecodexcoding-agentsdesktop-appdeveloper-toolselectrongemini-cliharnessharness-engineeringlocal-firstmemorymulti-agentopencodeorchestrationtypescript
blog: rewrite the install and first hour guides for 0.5.6, add every agent on our floor
chaitanyagiri306e778ab7bb
diff --git a/README.md b/README.mdindex fc1e934b..e94e2582 100644--- a/README.md+++ b/README.md@@ -30,7 +30,7 @@ visualized as avatars at work on a shared office floor.<p><a href="./LICENSE"><img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-F4D35E.svg?style=flat-square&labelColor=6E1423"></a>- <a href="./CHANGELOG.md"><img alt="Version: 0.5.3" src="https://img.shields.io/badge/version-0.5.3-F4D35E.svg?style=flat-square&labelColor=6E1423"></a>+ <a href="./CHANGELOG.md"><img alt="Version: 0.5.6" src="https://img.shields.io/badge/version-0.5.6-F4D35E.svg?style=flat-square&labelColor=6E1423"></a><a href="https://github.com/chaitanyagiri/munder-difflin/releases"><img alt="Downloads across all releases" src="https://img.shields.io/github/downloads/chaitanyagiri/munder-difflin/total?style=flat-square&label=downloads&color=F4D35E&labelColor=6E1423"></a><img alt="Status: pre-release" src="https://img.shields.io/badge/status-pre--release-F4F1EA.svg?style=flat-square&labelColor=6E1423"><img alt="Platform: macOS | Windows | Linux" src="https://img.shields.io/badge/platform-macOS%20%7C%20Windows%20%7C%20Linux-F4F1EA.svg?style=flat-square&labelColor=6E1423">@@ -48,14 +48,17 @@ visualized as avatars at work on a shared office floor.<br><br>-**[⬇ Download 0.5.3 for macOS, Windows or Linux](https://harnessmd.com/download)**+**[⬇ Download 0.5.6 for macOS, Windows or Linux](https://harnessmd.com/download)**<sub>macOS builds are signed and notarized. You do not need to build from source to use it.-The open source build is on the <a href="https://github.com/chaitanyagiri/munder-difflin/releases/tag/v0.5.3">GitHub releases page</a>.</sub>+The open source build is on the <a href="https://github.com/HarnessMD/munder-difflin/releases">GitHub releases page</a>.</sub>-**Cancel your Granola and Wispr Flow subscriptions.** 0.5.3 dictates into any app and transcribes your+**Cancel your Granola and Wispr Flow subscriptions.** It dictates into any app and transcribes yourmeetings, all running locally.+**New here?** Start with the [install guide](https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/), then+[your first hour](https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/), and see [every agent on our own floor](https://munderdiffl.in/blog/ai-agent-use-cases-our-office/).+<sub><b>Launch offer:</b> the Pro annual plan for $150 USD, adjusted for purchasing power in differentcountries and as low as $100 a year. <a href="https://app.harnessmd.com/console/license">Get Pro</a></sub>@@ -318,7 +321,28 @@ than sending you to a docs page.- **Prerequisites**: one Settings page showing which supporting tools (uv, git, Node, MemPalace, each agent CLI) you have, what each is for, and a button that asks Michael to install what is missing.> [!NOTE]-> **Status: v0.5.3. Cancel your Granola and Wispr Flow subscriptions.**+> **Status: v0.5.6. Fewer tokens, temps for one job, and one Slack bot for every channel.**+> These notes describe the 0.5.6 app on the [download page](https://harnessmd.com/download).+> **Fewer tokens by default**: Compact when context fills and Clear on a switch of work are on for every new install.+> The clear runs when an agent's work switches after a session of 60 minutes, and a `handoff.md` in the agent's folder+> carries the thread across. A new install has no hourly standup.+> **Temps**: any lasting agent can start a temp, a short lived agent for one job, not only the orchestrator. Wherever you+> pick who answers (Slack, a webhook, a schedule's Sends to, the Stapler's Send captures to), you can pick **A temp** or a+> temp that clones one of your agents. **Temps per agent** (0 to 4) and **Max temps at once** sit on the orchestrator's+> **Budget & breaker** tab.+> **Slack**: one bot can listen to **Every channel the bot is in**, a direct message to the bot is always answered, and+> replies stay in the thread. The Slack app needs four more scopes: `im:history`, `im:read`, `mpim:history`, `mpim:read`.+> Slack comes back by itself after a failed start.+> **Sign in and accounts in the app**: press **Open browser to sign in** and sign in there; the screen moves on by itself,+> or you paste the code into **Code from the browser**. **Sign out** and **Switch to Teams** sit in the profile menu.+> **Circuit breaker**: an agent that stopped as told is no longer constrained again, and the same warning is not sent twice+> inside ten minutes.+> **Fixes**: the window no longer hangs after hours, hooks work in Claude Code's auto mode, sign in works on a network with+> its own certificate, a message to Codex no longer waits for Enter, the Windows folder name field shows the name and not+> the whole path, the Dock click works the first time with the Stapler on, and Haiku 5.5 is in the model list.+> New to it? The [install guide](https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/) and [your first hour](https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/) cover every step.+>+> **v0.5.3: Cancel your Granola and Wispr Flow subscriptions.**> The Stapler does both, all running locally, and every agent gets easier to start, restart and steer.> **Dictate into any app**: hold Option on the Mac (Microphone and Accessibility permissions), or> Control+Alt+Space on Windows and Linux X11, talk, and let go. Wayland says why it cannot.@@ -361,7 +385,7 @@ than sending you to a docs page.> and it has been removed entirely.> macOS (signed & notarized), Windows, and Linux builds are on the> [download page](https://harnessmd.com/download). The open source build is on the-> [releases page](https://github.com/chaitanyagiri/munder-difflin/releases/tag/v0.5.3).+> [releases page](https://github.com/HarnessMD/munder-difflin/releases).<div align="right">(<a href="#munder-difflin">↑ back to top</a>)</div>@@ -370,9 +394,10 @@ than sending you to a docs page.### Download the app**Most people want this one.** Munder Difflin is free and open source. Signed and notarized macOS-builds of 0.5.3, plus Windows and Linux, are on the [download page](https://harnessmd.com/download).-The open source build is on the [latest GitHub release](https://github.com/chaitanyagiri/munder-difflin/releases/tag/v0.5.3). Install it,-open it, and the wizard takes you the rest of the way. You do not need Node, a toolchain, or this+builds of 0.5.6, plus Windows and Linux, are on the [download page](https://harnessmd.com/download).+The open source build is on the [GitHub releases page](https://github.com/HarnessMD/munder-difflin/releases). Install it,+open it, and the wizard takes you the rest of the way. The [install guide](https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/) walks through+every step and every fix. You do not need Node, a toolchain, or thisrepository.You do still need at least one agent CLI on your machine. From 0.5.3 a missing CLI shows as a card@@ -443,7 +468,9 @@ visual system.## Roadmap-Shipped through **v0.5.3**: local dictation into any app and meetings with both sides of the call,+Shipped through **v0.5.6**: fewer tokens by default with a handoff file across a cleared session, temps that any+agent can start and that can answer inbound messages, one Slack bot for every channel it is in, sign in and accounts+inside the app, a quieter circuit breaker, local dictation into any app and meetings with both sides of the call,with the model in the installer, the new agent cards in the free build and a new Pro sidebar, amissing CLI you install and sign in to from inside the app, Ask me on every Inbox, more than onefloor on one machine, ticket keys and a Tasks tab on every agent, redesigned Settings with inbound webhooks and custom secrets, Opus 5.5 as the default Claudediff --git a/blog/src/_data/media.json b/blog/src/_data/media.jsonindex 2069b76a..c90448e8 100644--- a/blog/src/_data/media.json+++ b/blog/src/_data/media.json@@ -1501,11 +1501,11 @@"youtube": []},"how-to-install-and-use-munder-difflin": {- "title": "How to Install and Use Munder Difflin: A Beginner's Guide",+ "title": "How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office","category": "guides","hero": {"file": "assets/media/how-to-install-and-use-munder-difflin/hero.png",- "alt": "The first Munder Difflin onboarding screen, asking whether you are technical or non-technical",+ "alt": "Title card on white. Large ink type reads Install Munder Difflin 0.5.6 under the line From zero to a working office, with a Beginner's guide pill, beside a shelf going up on two brackets with a spirit level on it, a screwdriver, a hex key and two spare screws.","prompt": "","status": "ready"},@@ -1605,6 +1605,18 @@"alt": "Munder Difflin in focus mode, showing a single agent's terminal at full width with the agent list on the left","prompt": "","status": "ready"+ },+ "autonomy-056": {+ "file": "assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png",+ "alt": "Screenshot of Munder Difflin Settings, Autonomy & Budgets, in light mode: Autonomous set to autonomous, Who can add agents set to me and the orchestrator, the circuit breaker switched on with fields for a floor token budget and token velocity, and archiving settings.",+ "prompt": "",+ "status": "ready"+ },+ "prerequisites-056": {+ "file": "assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png",+ "alt": "Screenshot of Munder Difflin Settings, Prerequisites, in light mode: uv, git and Node.js marked Found with their paths, and MemPalace marked Not found with its install command and a copy button.",+ "prompt": "",+ "status": "ready"}},"youtube": []@@ -3822,11 +3834,11 @@"youtube": []},"your-first-hour-with-munder-difflin": {- "title": "Your First Hour With Munder Difflin: Onboarding to a Pro Office",+ "title": "Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office","category": "guides","hero": {"file": "assets/media/your-first-hour-with-munder-difflin/hero.png",- "alt": "Michael, the orchestrator, in the Pro app after dispatching four tasks to his agents",+ "alt": "Title card on white. Large ink type reads Your First Hour With Munder Difflin 0.5.6, with the line From first sign in to a working office and a pill reading Onboarding guide, beside a blue stopwatch whose yellow ring has gone once all the way round as it rings.","prompt": "","status": "ready"},@@ -3986,6 +3998,18 @@"alt": "The floating Stapler with its action ring open","prompt": "","status": "ready"+ },+ "pro-sidebar-056": {+ "file": "assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png",+ "alt": "Screenshot of the Munder Difflin Pro sidebar in light mode: the app name and PRO plan, then rows for Tasks, Inbox, Automations, Memory, Capabilities and Stapler.",+ "prompt": "",+ "status": "ready"+ },+ "ask-me-056": {+ "file": "assets/media/your-first-hour-with-munder-difflin/ask-me-056.png",+ "alt": "Screenshot of an agent's Inbox in Munder Difflin, light mode: an Ask me list with three questions, each with a task key, a time and an Answer button, above the message box with Send and Steer.",+ "prompt": "",+ "status": "ready"}},"youtube": [@@ -4858,5 +4882,23 @@"status": "ready"},"inline": {}+ },+ "ai-agent-use-cases-our-office": {+ "title": "AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does",+ "category": "guides",+ "hero": {+ "file": "assets/media/ai-agent-use-cases-our-office/hero.png",+ "alt": "Title card on white. Every Agent on Our Floor in large ink type, a Checked 14 Oct 2026 pill and a yellow desk bell ringing.",+ "prompt": "Still of the hero scene in assets/media/ai-agent-use-cases-our-office/motion.js.",+ "status": "ready"+ },+ "inline": {+ "busy-office": {+ "file": "assets/media/ai-agent-use-cases-our-office/busy-office.png",+ "alt": "Screenshot of the middle of our real Munder Difflin office floor in the Classic view: seven agents on a green tiled floor among desks, each with a name label, Jim, Angela, Ryan, Stanley, Creed, Dwight and Kevin. Thought bubbles above them read idle and starting up.",+ "prompt": "",+ "status": "ready"+ }+ }}}diff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/busy-office.png b/blog/src/assets/media/ai-agent-use-cases-our-office/busy-office.pngnew file mode 100644index 00000000..f7e4974dBinary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/busy-office.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/clock.png b/blog/src/assets/media/ai-agent-use-cases-our-office/clock.pngnew file mode 100644index 00000000..e5c1d5e6Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/clock.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/drafts.png b/blog/src/assets/media/ai-agent-use-cases-our-office/drafts.pngnew file mode 100644index 00000000..2db27e53Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/drafts.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/floor.png b/blog/src/assets/media/ai-agent-use-cases-our-office/floor.pngnew file mode 100644index 00000000..19b4ff74Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/floor.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/gate.png b/blog/src/assets/media/ai-agent-use-cases-our-office/gate.pngnew file mode 100644index 00000000..51480748Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/gate.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/hero.png b/blog/src/assets/media/ai-agent-use-cases-our-office/hero.pngnew file mode 100644index 00000000..a2b4d4d3Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/hero.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/motion.css b/blog/src/assets/media/ai-agent-use-cases-our-office/motion.cssnew file mode 100644index 00000000..7658a3ff--- /dev/null+++ b/blog/src/assets/media/ai-agent-use-cases-our-office/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "Every Agent on Our Floor". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/motion.js b/blog/src/assets/media/ai-agent-use-cases-our-office/motion.jsnew file mode 100644index 00000000..b859951d--- /dev/null+++ b/blog/src/assets/media/ai-agent-use-cases-our-office/motion.js@@ -0,0 +1,619 @@+/* Bright coded scenes for "Every Agent on Our Floor". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.+ To add a scene: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } above the player. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };++ const SC = {};++ /* ---- floor: eleven desks seen from above; a name card drops on each and its plant grows a leaf ---- */+ SC.floor = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "What does the whole floor look like?", "tap a desk");+ const SEATS = [+ ["Orchestrator", "Orchestrator: takes your orders, splits them into jobs", YEL],+ ["Mission control", "Mission control: keeps one page of company numbers fresh", SKY],+ ["Community lead", "Community lead: a digest and a demand list", MINT],+ ["Growth", "Growth: researches and writes this blog", LIL],+ ["Motion graphics", "Motion graphics: launch videos and animation loops", BLUE],+ ["Research", "Research: a daily scout report, a link for every claim", MINT],+ ["Founder's office", "Founder's office: a morning briefing, Slack replies, first drafts", YEL],+ ["Scrum master", "Scrum master: keeps the task board readable", SKY],+ ["Engineering lead", "Engineering lead: plans the work and gates every piece", LIL],+ ["Engineers", "Engineers: build the pieces", BLUE],+ ["Designer", "Designer: draws every new screen before any code", MINT],+ ];+ const X0 = 36, PW = 182, Y0 = 100, PH = 131, DH = 92, ANG = [-58, 62, 180, 0, 122, -120];+ const A = SEATS.map((s, i) => 0.9 + i * 0.72);+ el("rect", { x: 26, y: 88, width: 748, height: 408, rx: 16, fill: SOFT }, svg);+ let tiles = "";+ for (let i = 1; i < 16; i++) tiles += `M${26 + i * 46.75} 90 V494 `;+ for (let i = 1; i < 8; i++) tiles += `M28 ${88 + i * 51} H772 `;+ el("path", { d: tiles, stroke: "#fff", "stroke-width": 2 }, svg);+ let pinned = -1;+ const desks = SEATS.map(([name, line, col], i) => {+ const wide = i === 9, c = i < 8 ? i % 4 : i === 8 ? 0 : i === 9 ? 1 : 3, r = i < 8 ? (i / 4) | 0 : 2;+ const x = X0 + c * PW + 6, y = Y0 + r * PH, w = (wide ? 2 : 1) * PW - 12, xs = wide ? [40, 112, 184, 256] : [40], px = w - 30, py = 27;+ const g = el("g", { transform: `translate(${x} ${y})` }, svg);+ el("rect", { x: -5, y: -5, width: w + 10, height: DH + 32, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const chairs = xs.map((cx) => {+ const ch = el("g", {}, g);+ el("rect", { x: -19, y: -14, width: 38, height: 28, rx: 9, fill: col, stroke: INK, "stroke-width": 2.5 }, ch);+ el("rect", { x: -15, y: 11, width: 30, height: 8, rx: 4, fill: INK }, ch);+ return { ch, cx };+ });+ el("rect", { width: w, height: DH, rx: 10, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ xs.forEach((lx) => {+ el("rect", { x: lx - 21, y: 12, width: 42, height: 30, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: lx - 21, y: 12, width: 42, height: 11, rx: 4, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${lx - 13} 30 H${lx + 13} M${lx - 13} 36 H${lx + 13}`, stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-dasharray": "1 5" }, g);+ });+ if (!wide) {+ el("rect", { x: 76, y: 13, width: 26, height: 30, rx: 3, fill: SOFT, stroke: INK, "stroke-width": 2.5, transform: `rotate(${(i % 3) * 5 - 5} 89 28)` }, g);+ el("path", { d: "M82 22 H96 M82 28 H96 M82 34 H91", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", transform: `rotate(${(i % 3) * 5 - 5} 89 28)` }, g);+ }+ el("circle", { cx: px, cy: py, r: 12, fill: YEL, stroke: INK, "stroke-width": 2.5 }, g);+ const leaves = ANG.map(() => el("path", { d: "M0 0 Q8 -10 0 -21 Q-8 -10 0 0 Z", fill: MINT, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, g));+ el("circle", { cx: px, cy: py, r: 3, fill: INK }, g);+ const card = el("g", {}, g);+ el("rect", { x: -79, y: -17, width: 158, height: 34, rx: 6, fill: col, stroke: INK, "stroke-width": 2.5 }, card);+ fit(txt(card, 0, 6.5, name, { "text-anchor": "middle", "font-size": 18, "font-weight": 700 }), 144);+ const d = { card, leaves, chairs, cx: w / 2, cy: DH - 23, px, py, kt: -99, extra: 0 };+ button(g, line, () => { d.extra = d.extra >= 3 ? 0 : d.extra + 1; d.kt = api.now(); pinned = i; api.poke(); });+ return d;+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), idle = io(seg(t, 9.2, 10.2)) * (1 - out);+ let cur = -1;+ desks.forEach((d, i) => {+ const a = A[i], k = seg(t, a, a + 0.42), e = cl(bk(k, 1.6), 0, 1.2), tau = t - a - 0.42, w = now - d.kt;+ const rot = (1 - oc(k)) * (i % 2 ? 16 : -16) + (tau > 0 ? 3 * Math.exp(-5 * tau) * Math.sin(16 * tau) : 0);+ d.card.setAttribute("transform", `translate(${d.cx} ${(d.cy - 34 * (1 - oc(k))).toFixed(2)}) rotate(${rot.toFixed(2)}) scale(${lerp(1.9, 1, e).toFixed(3)})`);+ d.card.setAttribute("opacity", (seg(t, a, a + 0.12) * (1 - out)).toFixed(2));+ d.leaves.forEach((lf, j) => {+ const g = j < 2 ? 1 : j === 2 ? cl(bk(seg(t, a + 0.35, a + 0.95), 2.4), 0, 1.5) * (1 - out) : j - 3 >= d.extra ? 0 : j - 3 === d.extra - 1 ? cl(bk(seg(w, 0, 0.6), 2.4), 0, 1.5) : 1;+ lf.setAttribute("transform", `translate(${d.px} ${d.py}) rotate(${(ANG[j] + Math.sin(now * 1.3 + i * 0.9 + j * 2.1) * 4).toFixed(2)}) scale(${g.toFixed(3)})`);+ });+ /* one chair is left spinning */+ d.chairs.forEach(({ ch, cx }) => ch.setAttribute("transform", `translate(${cx} ${(DH + 4 + (i === 5 ? 5 * idle : 0)).toFixed(2)}) rotate(${(i === 5 ? 22 * idle * Math.sin(now * 0.9 + 1.1) : 0).toFixed(2)})`));+ if (t >= a && !out) cur = i;+ });+ const pw = pinned >= 0 ? now - desks[pinned].kt : 99;+ say(cap, pw < 4 ? SEATS[pinned][1] : cur < 0 ? "Fourteen seats, and nobody shares a job" : t >= 9.4 ? "Fourteen agents, each with one job" : SEATS[cur][1], 700);+ };+ },+ };++ /* ---- clock: the hand sweeps one day; a card slides out from behind the clock for each schedule ---- */+ SC.clock = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Which agents work on a clock?", "tap the clock");+ const CX = 400, CY = 284, R = 128;+ const hourAt = (t) => { const p = seg(t, 1, 9.4); return 24 * p * p * (3 - 2 * p); };+ const timeOf = (H) => { let a = 1, b = 9.4; for (let i = 0; i < 28; i++) { const m = (a + b) / 2; if (hourAt(m) < H) a = m; else b = m; } return b; };+ const C = [+ [152, 154, "09:00", "briefing", YEL, [9], 0, "Andy, 09:00: one briefing as a draft"],+ [152, 420, "09:30 and 21:30", "blog runs", MINT, [9.5, 21.5], 0.3, "Jim, 09:30 and 21:30: one pull request per run"],+ [648, 420, "every 6 hours", "dashboard", SKY, [6, 12, 18, 24], 0, "Ryan, every 6 hours: the company dashboard"],+ [648, 154, "every 12 hours", "task board", LIL, [12, 24], 0, "Meredith, every 12 hours: keeps the task board readable"],+ ];+ C.forEach(([x, y]) => el("rect", { x: x - 118, y: y - 52, width: 236, height: 104, rx: 14, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg));+ const cards = C.map(([x, y, a, b, col, hrs, lag, line]) => {+ const g = el("g", {}, svg);+ el("rect", { x: -118, y: -52, width: 236, height: 104, rx: 14, fill: col, stroke: INK, "stroke-width": 3 }, g);+ fit(txt(g, -98, -9, a, { "font-size": 25, "font-weight": 700 }), 196);+ txt(g, -98, 27, b, { "font-size": 22, "font-weight": 500 });+ const pips = hrs.map((H, j) => ({ c: el("circle", { cx: 100 - (hrs.length - 1 - j) * 19, cy: 20, r: 6.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g), at: timeOf(H) + lag }));+ return { g, x, y, t0: timeOf(hrs[0]) - 0.6 + lag, pips, line };+ });+ const clock = el("g", {}, svg);+ el("circle", { cx: CX, cy: CY, r: R + 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, clock);+ const pen = el("g", {}, clock);+ el("path", { d: "M0 0 V52", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, pen);+ el("circle", { cy: 52, r: 15, fill: YEL, stroke: INK, "stroke-width": 3 }, pen);+ el("circle", { cx: CX, cy: CY - R - 14, r: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, clock);+ el("circle", { cx: CX, cy: CY, r: R + 11, fill: BLUE, stroke: INK, "stroke-width": 3 }, clock);+ el("circle", { cx: CX, cy: CY, r: R - 3, fill: "#fff", stroke: INK, "stroke-width": 3 }, clock);+ /* the tick at seven was put on by eye */+ for (let i = 0; i < 12; i++) el("path", { d: `M${CX} ${CY - R + 11} V${CY - R + (i % 3 ? 25 : 37)}`, stroke: INK, "stroke-width": i % 3 ? 3 : 6, "stroke-linecap": "round", transform: `rotate(${i * 30} ${CX} ${CY}) rotate(${i === 7 ? 16 : 0} ${CX} ${CY - R + 18})` }, clock);+ const hand = el("g", {}, clock);+ el("path", { d: `M0 18 V${-(R - 44)}`, stroke: INK, "stroke-width": 9, "stroke-linecap": "round" }, hand);+ el("circle", { cy: -(R - 44), r: 9, fill: YEL, stroke: INK, "stroke-width": 3 }, hand);+ el("circle", { cx: CX, cy: CY, r: 10, fill: YEL, stroke: INK, "stroke-width": 3 }, clock);+ button(clock, "Run the day again", () => api.seek(0.9));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const tau = t - 9.4, out = seg(t, 13.2, 13.7);+ const ang = hourAt(t) * 30 + (tau > 0 ? 6 * Math.exp(-4 * tau) * Math.sin(11 * tau) : 0);+ hand.setAttribute("transform", `translate(${CX} ${CY}) rotate(${ang.toFixed(2)})`);+ pen.setAttribute("transform", `translate(${CX} ${CY + R + 8}) rotate(${(Math.sin(now * 2.6 + 0.9) * 20).toFixed(2)})`);+ let cur = -1, last = -1;+ cards.forEach((c, i) => {+ const k = seg(t, c.t0, c.t0 + 0.6), e = bk(k, 1.5);+ let bump = 0;+ c.pips.forEach((p) => {+ const w = t - p.at, on = w >= 0 && out < 1;+ p.c.setAttribute("fill", on ? INK : "#fff");+ p.c.setAttribute("r", (6.5 * (on ? 1 + 0.7 * Math.exp(-7 * w) : 1)).toFixed(2));+ if (on) bump += Math.exp(-6 * w);+ });+ tf(c.g, lerp(CX, c.x, e).toFixed(2), lerp(CY, c.y, e).toFixed(2), (lerp(0.45, 1, cl(e, 0, 1.1)) * (1 + 0.06 * bump)).toFixed(4), ((1 - oc(k)) * (i % 2 ? 10 : -10)).toFixed(2));+ c.g.setAttribute("opacity", t >= c.t0 ? (1 - out).toFixed(2) : 0);+ if (t >= c.t0 && c.t0 > last) { last = c.t0; cur = i; }+ });+ say(cap, cur < 0 || out > 0 ? "Four of ours start from a schedule" : t > 9.8 ? "Four of ours start from a schedule, with no person typing anything" : cards[cur].line, 700);+ };+ },+ };++ /* ---- gate: a drawing becomes a jigsaw piece, waits at a raised drawbridge for three checks, then joins the row ---- */+ SC.gate = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Who writes the code, and who checks it?", "tap the bridge");+ const G = 400, PIECE = "M-20 -20 H20 V-6 h5 a8 8 0 1 1 0 12 h-5 V20 H-20 V6 h5 a8 8 0 1 0 0 -12 h-5 Z";+ const piece = (p, fill) => el("path", { d: PIECE, fill, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, p);+ const water = el("path", { fill: SKY, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 30, y: G, width: 446, height: 54, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 604, y: G, width: 166, height: 54, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 118, 435, "draws first", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 });+ txt(svg, 322, 435, "build to an order", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 });+ fit(txt(svg, 687, 435, "release branch", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }), 148);+ /* drawing board */+ el("path", { d: `M70 ${G} L104 236 M166 ${G} L132 236 M118 ${G} V250`, stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, svg);+ const board = el("g", { transform: "translate(118 232) rotate(-7)" }, svg);+ el("rect", { x: -72, y: -92, width: 144, height: 170, rx: 8, fill: LIL, stroke: INK, "stroke-width": 3 }, board);+ el("rect", { x: -78, y: 78, width: 156, height: 12, rx: 5, fill: YEL, stroke: INK, "stroke-width": 3 }, board);+ el("rect", { x: -54, y: -76, width: 108, height: 140, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, board);+ el("path", { d: "M-40 -54 H40 M-40 -30 H8 M-40 -6 H24", stroke: SOFT, "stroke-width": 8, "stroke-linecap": "round" }, board);+ el("path", { d: "M22 72 H58 L68 75 L58 78 H22 Z", fill: SKY, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, board);+ /* tool rail and workbench */+ el("path", { d: "M240 146 H404 M262 146 V158 M300 146 V158 M385 146 V158", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, svg);+ el("rect", { x: 254, y: 156, width: 16, height: 96, rx: 3, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M254 172 h7 M254 188 h7 M254 204 h7 M254 220 h7 M254 236 h7", stroke: INK, "stroke-width": 2 }, svg);+ el("path", { d: "M296 156 V236 H360 Z M308 192 V224 H334 Z", fill: SKY, "fill-rule": "evenodd", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 380, y: 156, width: 10, height: 76, rx: 4, fill: LIL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 385, cy: 240, r: 11, fill: LIL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 385, cy: 240, r: 4, fill: "#fff", stroke: INK, "stroke-width": 2 }, svg);+ [254, 378].forEach((x) => el("rect", { x, y: 344, width: 12, height: 56, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg));+ el("path", { d: "M266 376 H378", stroke: INK, "stroke-width": 3 }, svg);+ const bench = el("rect", { x: 238, y: 330, width: 170, height: 15, rx: 5, fill: YEL, stroke: INK, "stroke-width": 3 }, svg);+ const ham = el("g", {}, svg);+ el("rect", { x: -50, y: -3, width: 50, height: 6, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ham);+ el("rect", { x: -62, y: -13, width: 16, height: 22, rx: 3, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, ham);+ /* the gate: a sign with three lights on a post, and the bridge it holds up */+ el("rect", { x: 462, y: 248, width: 16, height: 152, fill: BLUE, stroke: INK, "stroke-width": 3 }, svg);+ /* a lifebuoy on the post, in case a piece falls in */+ el("circle", { cx: 470, cy: 318, r: 13, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 470, cy: 318, r: 5, fill: BLUE, stroke: INK, "stroke-width": 2 }, svg);+ el("path", { d: "M470 305 V313 M470 323 V331 M457 318 H465 M475 318 H483", stroke: INK, "stroke-width": 2 }, svg);+ el("rect", { x: 462, y: 98, width: 290, height: 150, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 486, 134, "the gate", { "font-size": 26, "font-weight": 700 });+ el("path", { d: "M486 148 H728", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, svg);+ const L = [5.3, 6.1, 6.9];+ const lamps = ["reads the diff", "own tests", "full suite"].map((s, i) => {+ txt(svg, 522, 178 + i * 29, s, { "font-size": 22, "font-weight": 600 });+ return el("circle", { cx: 498, cy: 171 + i * 29, r: 10, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ });+ const br = el("g", {}, svg);+ el("rect", { x: 468, y: 258, width: 146, height: 152, rx: 12, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, br);+ const rope = el("path", { stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-dasharray": "2 6" }, br);+ const deck = el("g", {}, br);+ el("rect", { x: 0, y: -8, width: 130, height: 11, rx: 4, fill: YEL, stroke: INK, "stroke-width": 3 }, deck);+ el("path", { d: "M26 -8 V3 M52 -8 V3 M78 -8 V3 M104 -8 V3", stroke: INK, "stroke-width": 2 }, deck);+ button(br, "Send another piece to the gate", () => api.seek(0.9));+ /* the release branch: a row of pieces with room for one more */+ [[684, BLUE], [724, LIL]].forEach(([x, c]) => piece(el("g", { transform: `translate(${x} ${G - 20})` }, svg), c));+ const sp = el("g", {}, svg); star(sp, 0, 0, 13, "#fff");+ /* the sheet that leaves the board, and the piece it becomes */+ const sheet = el("g", {}, svg);+ el("rect", { x: -54, y: -70, width: 108, height: 140, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sheet);+ const parts = [+ el("rect", { x: -42, y: -58, width: 84, height: 18, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2 }, sheet),+ el("rect", { x: -42, y: -30, width: 38, height: 44, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2 }, sheet),+ el("path", { d: "M6 -24 H42 M6 -9 H42 M6 6 H28", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, sheet),+ el("rect", { x: -42, y: 28, width: 46, height: 20, rx: 10, fill: MINT, stroke: INK, "stroke-width": 2 }, sheet),+ ];+ const pc = el("g", {}, svg); piece(pc, YEL);+ const HOPS = [[4.0, 4.7, 420, G - 20, 34], [8.8, 9.3, 506, G - 28, 22], [9.3, 9.8, 586, G - 28, 22], [9.8, 10.4, 644, G - 20, 28]];+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.1, 13.5);+ let w = `M476 456 V${(424 + 3 * Math.sin(now * 2.2)).toFixed(1)}`;+ for (let x = 492; x <= 604; x += 16) w += ` L${x} ${(424 + 3 * Math.sin(x * 0.11 + now * 2.2)).toFixed(1)}`;+ water.setAttribute("d", w + " V456 Z");+ /* sheet: drawn on the board, then flies to the bench */+ parts.forEach((p, i) => p.setAttribute("opacity", seg(t, 0.25 + i * 0.26, 0.5 + i * 0.26).toFixed(2)));+ const f = io(seg(t, 1.5, 2.6)), taps = t > 2.6 && t < 3.5 ? Math.abs(Math.sin(((t - 2.6) / 0.3) * Math.PI)) : 0;+ tf(sheet, lerp(117.3, 300, f).toFixed(2), (lerp(226, 296, f) - 76 * Math.sin(Math.PI * f) + (taps ? (1 - taps) * 3 : 0)).toFixed(2), lerp(1, 0.4, f).toFixed(4), (lerp(-7, 0, f) + 16 * Math.sin(Math.PI * f)).toFixed(2));+ sheet.setAttribute("opacity", (1 - seg(t, 3.15, 3.35)).toFixed(2));+ tf(ham, 402, 323, 1, (50 * taps).toFixed(2));+ bench.setAttribute("y", (330 + (taps && taps < 0.25 ? 1.5 : 0)).toFixed(1));+ /* the piece */+ let x = 300, y = 310, r = 0;+ HOPS.forEach(([a, b, X, Y, H]) => { const p = seg(t, a, b); if (p > 0) { const e = p < 1 ? p : 1; x = lerp(x, X, e); y = lerp(y, Y, e) - H * 4 * e * (1 - e); r = 9 * Math.sin(2 * Math.PI * e); } });+ const t1 = t - 4.7, t2 = t - 10.4;+ if (t1 > 0 && t < 8.8) x -= 6 * Math.exp(-7 * t1) * Math.abs(Math.sin(13 * t1));+ if (t2 > 0) r = 5 * Math.exp(-6 * t2) * Math.sin(18 * t2);+ tf(pc, x.toFixed(2), y.toFixed(2), cl(bk(seg(t, 3.2, 3.65), 2.2), 0, 1.4).toFixed(3), r.toFixed(2));+ pc.setAttribute("opacity", (1 - out).toFixed(2));+ lamps.forEach((c, i) => { const v = t - L[i], on = v >= 0 && t < 13.2; c.setAttribute("fill", on ? MINT : SOFT); c.setAttribute("r", (10 * (on ? 1 + 0.5 * Math.exp(-7 * v) : 1)).toFixed(2)); });+ const k = io(seg(t, 7.6, 8.5)) * (1 - io(seg(t, 13.2, 13.9))), tb = t - 8.5;+ const ang = -76 * (1 - k) - (tb > 0 && t < 13.2 ? 7 * Math.exp(-5 * tb) * Math.abs(Math.sin(11 * tb)) : 0), ar = (ang * Math.PI) / 180;+ deck.setAttribute("transform", `translate(478 398) rotate(${ang.toFixed(2)})`);+ rope.setAttribute("opacity", (1 - k).toFixed(2));+ rope.setAttribute("d", `M478 258 L${(478 + 124 * Math.cos(ar)).toFixed(1)} ${(394 + 124 * Math.sin(ar)).toFixed(1)}`);+ tf(sp, 646, 334, (cl(bk(seg(t, 10.4, 10.9), 2.6), 0, 1.5) * (1 - out) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.5 || out > 0 ? "Pam draws first: every new screen is an HTML page" : t < 4 ? "Builders build to an order" : t < 5.3 ? "Kevin is the gate" : t < 10.4 ? "He reads the diff, writes his own tests, runs the full suite" : "He merges into a release branch, never into main", 700);+ };+ },+ };++ /* ---- title card, 16 by 9: a desk bell that somebody keeps ringing ---- */+ SC.hero = {+ w: 960, h: 540, dur: 3.2, still: 0.55,+ build(svg) {+ txt(svg, 66, 112, "AI agent use cases from a real company", { "font-size": 25, "font-weight": 500, fill: MUTE });+ txt(svg, 62, 204, "Every Agent", { "font-size": 82, "font-weight": 700 });+ const ti = txt(svg, 62, 294, "on Our Floor", { "font-size": 82, "font-weight": 700 });+ el("path", { d: `M68 324 H${Math.round(56 + ti.getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ el("rect", { x: 62, y: 362, width: 312, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 218, 399, "Checked 14 Oct 2026", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 });+ el("path", { d: "M70 478 C190 440 300 508 420 468 S560 470 620 452", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 478, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ const c = el("g", {}, svg), bell = el("g", {}, c);+ el("rect", { x: -7, y: -92, width: 14, height: 40, fill: "#fff", stroke: INK, "stroke-width": 3 }, bell);+ const knob = el("rect", { x: -24, y: -112, width: 48, height: 20, rx: 10, fill: LIL, stroke: INK, "stroke-width": 3 }, bell);+ el("path", { d: "M-104 72 C-104 -22 -58 -64 0 -64 C58 -64 104 -22 104 72 Z", fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, bell);+ el("path", { d: "M-70 26 C-66 -10 -44 -32 -16 -40", stroke: "#fff", "stroke-width": 9, fill: "none", "stroke-linecap": "round" }, bell);+ el("rect", { x: -124, y: 72, width: 248, height: 24, rx: 12, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ el("path", { d: "M-150 110 H150", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, c);+ const ding = el("path", { d: "M122 -46 L152 -63 M134 -4 L164 -7 M92 -86 L110 -116 M-122 -46 L-152 -63 M-134 -4 L-164 -7 M-92 -86 L-110 -116", stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, c);+ star(svg, 626, 132, 15, SKY); star(svg, 880, 468, 17, MINT); star(svg, 596, 396, 12, YEL);+ return (t) => {+ const press = Math.sin(Math.PI * seg(t, 0.1, 0.5)), tau = t - 0.3;+ knob.setAttribute("y", (-112 + 16 * press).toFixed(2));+ bell.setAttribute("transform", `rotate(${(tau > 0 ? 2.4 * Math.exp(-2.4 * tau) * Math.sin(26 * tau) : 0).toFixed(3)} 0 72)`);+ ding.setAttribute("opacity", (seg(t, 0.3, 0.42) * (1 - seg(t, 1.2, 2))).toFixed(2));+ tf(ding, 0, 0, (1 + 0.1 * oc(seg(t, 0.3, 1.2))).toFixed(3));+ tf(c, 760, 312, 0.96);+ };+ },+ };++ /* a status pill sized to its word, with or without a tick; used by the scenes below */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);++ /* ---- drafts: three clipboards; a pen signs the first two, the third only ever collects labels ---- */+ SC.drafts = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Who deals with the outside world?", "tap the pen");+ const X = [146, 400, 654], TOP = 92, COL = [YEL, SKY, LIL], NAME = ["email draft", "pull request", "triage label"], WORD = ["sent", "merged", "labels only"], BT = [4.85, 7.45, 9.1];+ const SIG = "M-56 240 C-52 220 -40 220 -42 236 C-44 248 -30 246 -26 232 C-22 220 -12 222 -14 234 C-16 246 -2 244 2 232 C6 220 16 222 14 234 C14 244 26 244 32 232 C38 220 46 236 58 230";+ const TAGS = [[-18, 152, -4, MINT, 1.7], [14, 180, 4, YEL, 2.2], [-14, 208, 2, SKY, 8.1], [20, 236, -5, BLUE, 8.6]];+ const tag = (p, col) => {+ const g = el("g", {}, p);+ el("rect", { x: -38, y: -14, width: 76, height: 28, rx: 14, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ el("circle", { cx: -26, cy: 0, r: 4.5, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ return g;+ };+ /* the desk edge the pen lies on */+ [344, 432].forEach((x) => el("path", { d: `M${x - 13} 432 V442 a13 13 0 0 0 26 0 V432`, fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg));+ el("rect", { x: 30, y: 452, width: 740, height: 40, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 400, 479, "a person", { "text-anchor": "middle", "font-size": 20, "font-weight": 700 });+ const boards = X.map((x, i) => {+ const g = el("g", { transform: `translate(${x} ${TOP})` }, svg);+ el("rect", { x: -102, y: 0, width: 204, height: 300, rx: 14, fill: COL[i], stroke: INK, "stroke-width": 3 }, g);+ const sh = el("g", {}, g), b = { sh, tags: [] };+ el("rect", { x: -86, y: 16, width: 172, height: 268, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sh);+ txt(sh, 0, 70, NAME[i], { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ const faint = (d, col = LIL) => el("path", { d, stroke: col, "stroke-opacity": col === LIL ? 0.45 : 1, "stroke-width": 9, "stroke-linecap": "round" }, sh);+ if (i === 0) faint("M-64 100 H-8 M-64 124 H64 M-64 148 H64 M-64 172 H22");+ if (i === 1) { faint("M-64 100 H20 M-64 172 H48"); faint("M-64 124 H64 M-64 148 H30", MINT); }+ if (i === 2) {+ faint("M-38 100 H58 M-64 124 H30");+ el("circle", { cx: -57, cy: 100, r: 8, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sh);+ b.tags = TAGS.map(([tx, ty, r, col, at]) => ({ g: tag(sh, col), tx, ty, r, at }));+ b.dang = tag(sh, MINT);+ } else {+ b.line = el("path", { d: "M-62 252 H62", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "1 8" }, sh);+ b.sig = el("path", { d: SIG, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, sh);+ }+ el("rect", { x: -40, y: -10, width: 80, height: 34, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("circle", { cx: 0, cy: 4, r: 5, fill: INK }, g);+ b.badge = pill(g, WORD[i], i === 2 ? YEL : MINT, i < 2);+ return b;+ });+ const SL = boards[0].sig.getTotalLength();+ boards.forEach((b) => b.sig && b.sig.setAttribute("stroke-dasharray", `${SL} ${SL}`));+ const at = (i, k) => { const p = boards[i].sig.getPointAtLength(cl(k) * SL); return [X[i] + p.x, TOP + p.y]; };+ const HOME = [322, 438], A0 = at(0, 0), A1 = at(0, 1), B0 = at(1, 0), B1 = at(1, 1), TILT = -55;+ const wr = (k) => lerp(k, io(k), 0.6);+ const pen = el("g", {}, svg);+ el("rect", { x: -10, y: -24, width: 156, height: 48, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, pen);+ el("rect", { x: 40, y: -12, width: 92, height: 24, rx: 11, fill: BLUE, stroke: INK, "stroke-width": 3 }, pen);+ el("path", { d: "M92 -12 V12", stroke: INK, "stroke-width": 2.5 }, pen);+ el("rect", { x: 92, y: -19, width: 36, height: 7, rx: 3.5, fill: YEL, stroke: INK, "stroke-width": 2 }, pen);+ el("rect", { x: 30, y: -11, width: 18, height: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, pen);+ el("path", { d: "M0 0 L30 -10 V10 Z", fill: YEL, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, pen);+ el("path", { d: "M8 0 H24", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, pen);+ button(pen, "Sign the drafts again", () => api.seek(0.3));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7);+ boards.forEach((b, i) => {+ const a = 0.25 + i * 0.35, k = seg(t, a, a + 0.45) * (1 - io(out)), v = t - a - 0.45;+ tf(b.sh, 0, (-24 * (1 - oc(k))).toFixed(2), 1, ((1 - oc(k)) * (i % 2 ? 6 : -6) + (v > 0 ? 2 * Math.exp(-5 * v) * Math.sin(14 * v) : 0)).toFixed(2));+ if (b.sig) {+ const s = wr(seg(t, i ? 5.5 : 3.8, i ? 6.4 : 4.7));+ b.sig.setAttribute("stroke-dashoffset", (SL * (1 - s)).toFixed(1));+ b.sig.setAttribute("opacity", s > 0 ? (1 - out).toFixed(2) : 0);+ b.line.setAttribute("opacity", s > 0 ? 1 : (0.6 + 0.4 * Math.sin(now * 4 + i * 2)).toFixed(2));+ }+ b.tags.forEach((g) => tf(g.g, g.tx, g.ty, (land(seg(t, g.at, g.at + 0.4), 2.4) * (1 - out)).toFixed(3), g.r));+ if (b.dang) {+ /* one label too many: it lets go and hangs off the edge */+ const w = t - 10, ang = w < 0 ? 172 : 98 + 74 * Math.exp(-2.2 * w) * Math.cos(5.5 * w) + 4 * Math.sin(now * 1.8) * seg(w, 1, 2);+ b.dang.setAttribute("transform", `translate(-76 150) rotate(${ang.toFixed(2)}) scale(${(land(seg(t, 9.8, 10.1), 2) * (1 - out)).toFixed(3)}) translate(26 0)`);+ }+ const bk2 = seg(t, BT[i], BT[i] + 0.45);+ tf(b.badge, 0, 300, land(bk2).toFixed(3), ((1 - oc(bk2)) * -10).toFixed(2));+ b.badge.setAttribute("opacity", bk2 > 0 ? (1 - out).toFixed(2) : 0);+ });+ let p = HOME, r = 0;+ if (t >= 3 && t < 3.8) { const k = io(seg(t, 3, 3.8)); p = [lerp(HOME[0], A0[0], k), lerp(HOME[1], A0[1], k) - 22 * Math.sin(Math.PI * k)]; r = TILT * k; }+ else if (t >= 3.8 && t < 4.7) { p = at(0, wr(seg(t, 3.8, 4.7))); r = TILT + 3 * Math.sin(t * 16); }+ else if (t >= 4.7 && t < 5.5) { const k = io(seg(t, 4.7, 5.5)); p = [lerp(A1[0], B0[0], k), lerp(A1[1], B0[1], k) - 18 * Math.sin(Math.PI * k)]; r = TILT; }+ else if (t >= 5.5 && t < 6.4) { p = at(1, wr(seg(t, 5.5, 6.4))); r = TILT + 3 * Math.sin(t * 16); }+ else if (t >= 6.4 && t < 7.4) { const k = io(seg(t, 6.55, 7.4)); p = [lerp(B1[0], HOME[0], k), lerp(B1[1] - 8 * oc(seg(t, 6.4, 6.6)), HOME[1], k)]; r = TILT * (1 - k); }+ else if (t >= 7.4) { const v = t - 7.4; r = -3.5 * Math.exp(-6 * v) * Math.sin(17 * v); }+ tf(pen, p[0].toFixed(2), p[1].toFixed(2), 1, r.toFixed(2));+ say(cap, t < 1.6 || out > 0 ? "Four agents read what is outside the office" : t < 3 ? "None of them takes the final step alone" : t < 5.4 ? "Replies to email are drafts" : t < 8 ? "Jim opens pull requests to this website. A person merges them." : "Angela can add labels. She cannot close or merge.", 700);+ };+ },+ };++ /* ---- temps: a tent is pitched for each job, reports done and is packed away; the house stays ---- */+ SC.temps = {+ dur: 15.2, still: 11.3,+ build(svg, api) {+ head(svg, "Not every job deserves a desk", "tap a job");+ const SX = [118, 290, 462], G = 440, COL = [SKY, LIL, YEL], T = [0.4, 4.1, 7.8], TENT = "M0 -165 L75 0 H-75 Z";+ const JOB = [["run the", "test suite"], ["fix one", "release bug"], ["free", "disk space"]];+ const LINE = ["Run the test suite and write an honest report", "Fix one release bug", "Free disk space on a laptop"];+ el("rect", { x: 30, y: G, width: 740, height: 46, rx: 8, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ /* the house, with its chimney going */+ const smoke = [0, 1, 2].map(() => el("circle", { fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 700, y: 196, width: 24, height: 60, fill: LIL, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M552 268 L656 176 L760 268 Z", fill: BLUE, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 570, y: 268, width: 172, height: 172, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 578, y: 282, width: 156, height: 40, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ fit(txt(svg, 656, 309, "lasting agent", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }), 140);+ el("path", { d: "M634 440 V372 a22 22 0 0 1 44 0 V440 Z", fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("circle", { cx: 668, cy: 402, r: 3.5, fill: INK }, svg);+ [584, 692].forEach((x) => {+ el("rect", { x, y: 358, width: 34, height: 40, rx: 4, fill: SKY, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: `M${x + 17} 358 V398 M${x} 378 H${x + 34}`, stroke: INK, "stroke-width": 2 }, svg);+ });+ const spots = SX.map((x, i) => {+ el("path", { d: `M${x} 186 V${G}`, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: TENT, transform: `translate(${x} ${G})`, fill: SOFT, stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg);+ const tk = el("g", { transform: `translate(${x} 150)` }, svg);+ el("rect", { x: -81, y: -46, width: 162, height: 126, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, tk);+ el("rect", { x: -75, y: -40, width: 150, height: 76, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, tk);+ el("circle", { cx: 0, cy: -40, r: 6, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, tk);+ JOB[i].forEach((s, j) => txt(tk, 0, -7 + j * 25, s, { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }));+ el("rect", { x: -50, y: 50, width: 100, height: 14, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, tk);+ const bar = el("rect", { x: -47, y: 53, width: 0, height: 8, rx: 4, fill: MINT }, tk), done = pill(tk, "done", MINT, true);+ button(tk, `Start a temp to ${JOB[i].join(" ")}`, () => api.seek(T[i]));+ return { x, bar, done };+ });+ spots.forEach((s, i) => {+ const tent = (s.tent = el("g", {}, svg));+ el("path", { d: "M-46 -64 L-82 0 M46 -64 L82 0", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, tent);+ el("path", { d: TENT, fill: COL[i], stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, tent);+ el("path", { d: "M0 -112 L26 0 H-26 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, tent);+ el("path", { d: "M0 -112 V0", stroke: INK, "stroke-width": 2 }, tent);+ el("circle", { cx: 0, cy: -165, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, tent);+ });+ spots.forEach((s, i) => {+ const bag = (s.bag = el("g", {}, svg));+ el("path", { d: "M-12 -30 Q0 -47 12 -30", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, bag);+ el("rect", { x: -36, y: -30, width: 72, height: 30, rx: 13, fill: COL[i], stroke: INK, "stroke-width": 3 }, bag);+ el("path", { d: "M-14 -30 V0 M14 -30 V0", stroke: INK, "stroke-width": 2.5 }, bag);+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 14.4, 14.9);+ smoke.forEach((c, j) => {+ const p = (now * 0.3 + j / 3) % 1;+ c.setAttribute("cx", (712 + 9 * Math.sin(p * 5 + j * 2)).toFixed(1)); c.setAttribute("cy", (186 - 70 * p).toFixed(1));+ c.setAttribute("r", (5 + 9 * p).toFixed(1)); c.setAttribute("opacity", (cl(p * 6) * (1 - p * p)).toFixed(2));+ });+ let cur = -1, line = "";+ spots.forEach((s, i) => {+ /* the third job took three goes, so its tent does too */+ const u = t - T[i], D = i === 2 ? 1.45 : 0.45, w0 = 0.7 + D, w1 = w0 + 0.9, f0 = w1 + 0.6, l0 = f0 + 0.3, l1 = l0 + 0.7;+ const kin = oc(seg(u, 0, 0.7)), ko = seg(u, l0, l1), early = u < w0;+ const bx = early ? lerp(-50, s.x, kin) : lerp(s.x, -50, ko * ko * ko);+ const hop = early ? 12 * Math.abs(Math.sin(kin * Math.PI * 3)) * (1 - kin) : 0;+ const bs = u <= 0 || u >= l1 ? 0 : early ? 1 - seg(u, 0.7, 0.85) : land(seg(u, f0 + 0.1, l0), 2);+ tf(s.bag, bx.toFixed(2), (G - hop).toFixed(2), bs.toFixed(3), (early ? 8 * Math.sin(kin * Math.PI * 3) * (1 - kin) : -9 * ko).toFixed(2));+ let h = 0;+ if (u >= 0.7 && u < f0) {+ const a = 0.7;+ if (i < 2) h = bk(seg(u, a, a + D), 1.8);+ else h = u < a + 0.3 ? 0.72 * oc(seg(u, a, a + 0.3)) : u < a + 0.55 ? lerp(0.72, 0.22, io(seg(u, a + 0.3, a + 0.55))) : u < a + 0.85 ? lerp(0.22, 0.86, oc(seg(u, a + 0.55, a + 0.85))) : u < a + 1.05 ? lerp(0.86, 0.28, io(seg(u, a + 0.85, a + 1.05))) : lerp(0.28, 1, bk(seg(u, a + 1.05, a + D), 1.8));+ } else if (u >= f0 && u < f0 + 0.3) { const k = seg(u, f0, f0 + 0.3); h = 1 - k * k; }+ const jig = u > w0 && u < w1 ? 1.1 * Math.sin(u * 30) : 0, joy = u >= w1 && u < w1 + 0.35 ? 9 * Math.sin(Math.PI * seg(u, w1, w1 + 0.35)) : 0;+ s.tent.setAttribute("transform", `translate(${s.x} ${(G - joy).toFixed(2)}) rotate(${jig.toFixed(2)}) scale(${(1 + 0.2 * (1 - cl(h))).toFixed(3)} ${Math.max(h, 0.001).toFixed(3)})`);+ s.tent.setAttribute("opacity", h > 0.01 ? 1 : 0);+ s.bar.setAttribute("width", (94 * (u >= w1 ? 1 : io(seg(u, w0, w1))) * (1 - out)).toFixed(1));+ tf(s.done, 0, 57, (land(seg(u, w1, w1 + 0.4)) * (1 - out)).toFixed(3));+ if (u >= 0) { cur = i; line = u < w1 ? LINE[i] : u < l1 ? "Removed when it reports done" : ""; }+ });+ say(cap, cur < 0 || out > 0 || (cur < 2 && !line) ? "A temp is started for one job" : line || "A temp is started for one job and removed when it reports done", 700);+ };+ },+ };++ /* ---- relay: a capsule goes outbox, the app, inbox through a tube; "done" comes back and the outboxes shut ---- */+ SC.relay = {+ dur: 13.5, still: 11.4,+ build(svg, api) {+ head(svg, "How does work move between them?", "tap the outbox");+ const P1 = "M96 200 H236 C300 200 292 134 340 134 C372 134 370 200 370 252 L430 348 C430 404 480 410 564 410 H704";+ const P2 = "M704 200 H564 C500 200 508 134 460 134 C428 134 430 200 430 252 L370 348 C370 404 320 410 236 410 H96";+ [[135, LIL, 12], [665, SKY, 88]].forEach(([cx, col, tab]) => {+ el("rect", { x: cx - 95 + tab, y: 104, width: 90, height: 40, rx: 9, fill: col, stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: cx - 95, y: 124, width: 190, height: 356, rx: 16, fill: col, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, cx, 250, "outbox", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ txt(svg, cx, 460, "inbox", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ /* a drawer for the memory file */+ el("rect", { x: cx - 62, y: 280, width: 124, height: 70, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: `M${cx - 16} 315 H${cx + 16}`, stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, svg);+ });+ const tube = (d) => {+ el("path", { d, fill: "none", stroke: INK, "stroke-width": 38, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ return el("path", { d, fill: "none", stroke: "#fff", "stroke-width": 32, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ };+ const t1 = tube(P1), t2 = tube(P2), L1 = t1.getTotalLength(), L2 = t2.getTotalLength();+ const capsule = (col, word) => {+ const g = el("g", {}, svg), w = word ? 66 : 52;+ el("rect", { x: -w / 2, y: -13, width: w, height: 26, rx: 13, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ if (word) txt(g, 0, 5.5, word, { "text-anchor": "middle", "font-size": 16, "font-weight": 700 });+ else {+ el("rect", { x: -11, y: -8, width: 22, height: 16, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-6 -2 H6 M-6 3 H2", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ }+ return g;+ };+ const c1 = capsule(BLUE), c2 = capsule(MINT, "done");+ [[230, 200], [570, 200], [230, 410], [570, 410]].forEach(([x, y]) => el("rect", { x: x - 6, y: y - 23, width: 12, height: 46, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, svg));+ const flap = () => {+ const g = el("g", {}, svg);+ el("rect", { x: -5, y: 0, width: 10, height: 38, rx: 4, fill: INK }, g);+ el("circle", { r: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, g);+ return g;+ };+ const v1 = flap(), v2 = flap();+ const app = el("g", {}, svg);+ el("rect", { x: -84, y: -60, width: 168, height: 120, rx: 16, fill: YEL, stroke: INK, "stroke-width": 3 }, app);+ txt(app, 0, -6, "the app", { "text-anchor": "middle", "font-size": 26, "font-weight": 700 });+ const light = el("circle", { cx: 0, cy: 28, r: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, app);+ const tick = el("path", { d: "M-6 28 l4.5 5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, app);+ const sp = el("g", {}, svg); star(sp, 0, 0, 15, "#fff");+ const file = el("g", {}, svg);+ el("path", { d: "M-15 -19 H7 L15 -11 V19 H-15 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, file);+ el("path", { d: "M7 -19 V-11 H15 M-8 -3 H8 M-8 5 H4", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-linejoin": "round" }, file);+ const hit = el("g", {}, svg);+ el("rect", { x: 66, y: 172, width: 176, height: 90, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Send a note from the outbox", () => api.seek(0.3));+ const put = (g, path, L, s, flip, sc) => {+ const p = path.getPointAtLength(s), a = path.getPointAtLength(Math.max(0, s - 1)), b = path.getPointAtLength(Math.min(L, s + 1));+ tf(g, p.x.toFixed(2), p.y.toFixed(2), sc.toFixed(3), ((Math.atan2(b.y - a.y, b.x - a.x) * 180) / Math.PI + (flip ? 180 : 0)).toFixed(2));+ return p;+ };+ const near = (p, on) => (on && Math.abs(p.x - 400) < 80 ? cl(1 - Math.abs(p.y - 300) / 75) : 0);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 12.7, 13.2), vis = 1 - out;+ /* a small file drops into the outbox and becomes a capsule */+ const fk = seg(t, 0.9, 1.4);+ tf(file, 135, lerp(100, 200, fk * fk).toFixed(2), (land(seg(t, 0.4, 0.75), 2) * lerp(1, 0.55, fk)).toFixed(3), (-8 * Math.sin(Math.PI * fk)).toFixed(2));+ file.setAttribute("opacity", (1 - seg(t, 1.3, 1.42)).toFixed(2));+ const s1 = t < 2.4 ? 40 - 9 * Math.sin(Math.PI * seg(t, 2.05, 2.4)) : lerp(40, L1 - 40, bk(seg(t, 2.4, 4.3), 0.8));+ const s2 = t < 6.6 ? 40 - 9 * Math.sin(Math.PI * seg(t, 6.25, 6.6)) : lerp(40, L2 - 40, bk(seg(t, 6.6, 8.3), 0.8));+ const a = put(c1, t1, L1, s1, false, land(seg(t, 1.35, 1.75), 2)), b = put(c2, t2, L2, s2, true, land(seg(t, 5.6, 6), 2));+ c1.setAttribute("opacity", vis.toFixed(2)); c2.setAttribute("opacity", vis.toFixed(2));+ const pulse = Math.max(near(a, t > 2.4 && t < 4.3), near(b, t > 6.6 && t < 8.3)), shut = seg(t, 8.8, 9.2) * vis;+ tf(app, 400, 300, (1 + 0.06 * pulse).toFixed(4), (1.6 * pulse * Math.sin(t * 40)).toFixed(2));+ light.setAttribute("fill", pulse > 0.1 || shut > 0 ? MINT : "#fff");+ light.setAttribute("r", (12 * (1 + 0.35 * Math.sin(Math.PI * shut))).toFixed(2));+ tick.setAttribute("opacity", shut.toFixed(2));+ /* the thread is closed: a flap drops over each outbox */+ const vk = cl(bk(seg(t, 9, 9.5), 1.4), 0, 1.3) * (1 - io(out));+ v1.setAttribute("transform", `translate(204 181) rotate(${(180 - 180 * vk).toFixed(2)})`);+ v2.setAttribute("transform", `translate(596 181) rotate(${(180 * vk - 180).toFixed(2)})`);+ tf(sp, 400, 440, (land(seg(t, 9.4, 9.9), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 2.4 || out > 0 ? "An agent drops a small file in its outbox" : t < 5.4 ? "The app delivers it" : t < 8.8 ? "Two kinds of message, inform and done, end a thread" : "Two agents cannot reply to each other forever", 700);+ };+ },+ };++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/relay.png b/blog/src/assets/media/ai-agent-use-cases-our-office/relay.pngnew file mode 100644index 00000000..874e066eBinary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/relay.png differdiff --git a/blog/src/assets/media/ai-agent-use-cases-our-office/temps.png b/blog/src/assets/media/ai-agent-use-cases-our-office/temps.pngnew file mode 100644index 00000000..5d18d488Binary files /dev/null and b/blog/src/assets/media/ai-agent-use-cases-our-office/temps.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.pngnew file mode 100644index 00000000..40a8464fBinary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/download.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/download.pngnew file mode 100644index 00000000..c94795a4Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/download.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/engine.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/engine.pngnew file mode 100644index 00000000..f3937b0aBinary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/engine.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/fixes.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/fixes.pngnew file mode 100644index 00000000..057bc795Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/fixes.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/fork.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/fork.pngnew file mode 100644index 00000000..28ecd4deBinary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/fork.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/hero.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/hero.pngindex fc51d30b..2ae7ec04 100644Binary files a/blog/src/assets/media/how-to-install-and-use-munder-difflin/hero.png and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/hero.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/ide.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/ide.pngnew file mode 100644index 00000000..39392202Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/ide.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.css b/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.cssnew file mode 100644index 00000000..859374ca--- /dev/null+++ b/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.js b/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.jsnew file mode 100644index 00000000..ee26de35--- /dev/null+++ b/blog/src/assets/media/how-to-install-and-use-munder-difflin/motion.js@@ -0,0 +1,1055 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.++ TO ADD A SCENE: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } just above+ the line marked "ADD NEW SCENES ABOVE THIS LINE". Change nothing else in this file.+ - The drawing is 800 by 600 units (shown 1600 by 1200). dur is the loop in seconds, still is the frame for NAME.png.+ - build draws everything once and returns the frame function: t is the loop time, now never wraps (idle motion).+ - api.seek(s) jumps the loop to second s (a replay). api.now() is the clock. api.poke() redraws after a tap.+ - Helpers: el, txt, head (heading plus tap hint), strip and say (the caption line), button (tap and keyboard),+ pill, star, fit, mv (marks a group that moves, for the measurement tests), seg, io, oc, bk, land, lerp, tf. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };+ /* a status pill sized to its word, with or without a tick */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* mark a group as moving, so the measurement tests can see when it crosses a label */+ const mv = (g) => { g.setAttribute("data-mv", ""); return g; };++ const SC = {};++ /* ---- title card, 16 by 9: a shelf going up, with a spirit level on it and a screw left over ---- */+ SC.hero = {+ w: 960, h: 540, dur: 4, still: 2.6,+ build(svg) {+ txt(svg, 66, 92, "From zero to a working office", { "font-size": 25, "font-weight": 500, fill: MUTE });+ const L = ["Install", "Munder Difflin", "0.5.6"].map((s, i) => txt(svg, 62, 170 + i * 78, s, { "font-size": 76, "font-weight": 700 }));+ const w = L[1].getComputedTextLength();+ if (w > 500) L.forEach((n) => n.setAttribute("font-size", Math.floor((76 * 500) / w)));+ el("path", { d: `M68 356 H${Math.round(56 + L[2].getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ const pt = txt(svg, 0, 427, "Beginner's guide", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 }), pw = Math.round(pt.getComputedTextLength() + 64);+ svg.insertBefore(el("rect", { x: 62, y: 390, width: pw, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }), pt);+ pt.setAttribute("x", 62 + pw / 2);+ el("path", { d: "M70 494 C190 468 300 518 420 490 S560 492 620 478", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 494, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ const c = el("g", { transform: "translate(748 300)" }, svg);+ /* the pencil line on the wall, the brackets and the plank */+ el("path", { d: "M-168 40 H168", stroke: LIL, "stroke-width": 4, "stroke-linecap": "round", "stroke-dasharray": "1 13" }, c);+ [-1, 1].forEach((s) => {+ el("path", { d: `M${s * 104} 66 V140 L${s * 34} 66 Z`, fill: LIL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, c);+ el("circle", { cx: s * 93, cy: 84, r: 4, fill: INK }, c);+ el("circle", { cx: s * 93, cy: 108, r: 4, fill: INK }, c);+ });+ el("rect", { x: -156, y: 42, width: 312, height: 26, rx: 7, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ /* the spirit level */+ const lv = el("g", { transform: "translate(-10 0)" }, c);+ el("rect", { x: -124, y: -16, width: 236, height: 58, rx: 10, fill: YEL, stroke: INK, "stroke-width": 3 }, lv);+ el("path", { d: "M-100 -16 V42 M88 -16 V42", stroke: INK, "stroke-width": 3 }, lv);+ el("rect", { x: -48, y: 0, width: 84, height: 26, rx: 13, fill: "#fff", stroke: INK, "stroke-width": 3 }, lv);+ const bubble = el("ellipse", { cy: 13, rx: 10, ry: 8, fill: MINT, stroke: INK, "stroke-width": 2.5 }, lv);+ el("path", { d: "M-19 0 V26 M7 0 V26", stroke: INK, "stroke-width": 2.5 }, lv);+ [-76, 62].forEach((x) => { el("circle", { cx: x, cy: 13, r: 10, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, lv); el("circle", { cx: x, cy: 13, r: 3.5, fill: MINT, stroke: INK, "stroke-width": 2 }, lv); });+ /* the screws that were left over, and the hex key that came in the box */+ const screw = (x, y, r) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, c);+ el("rect", { x: -11, y: -5, width: 30, height: 10, rx: 2, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-3 -5 V5 M4 -5 V5 M11 -5 V5", stroke: INK, "stroke-width": 2 }, g);+ el("rect", { x: -19, y: -10, width: 8, height: 20, rx: 3, fill: INK }, g);+ };+ screw(-74, -22, 0); screw(-128, -62, -38);+ const key = el("g", { transform: "translate(-62 -112) rotate(-16)" }, c);+ el("path", { d: "M-34 -46 V26 H36", fill: "none", stroke: INK, "stroke-width": 17, "stroke-linecap": "round", "stroke-linejoin": "round" }, key);+ el("path", { d: "M-34 -46 V26 H36", fill: "none", stroke: SKY, "stroke-width": 11, "stroke-linecap": "round", "stroke-linejoin": "round" }, key);+ /* the screwdriver, on the last screw */+ const drv = el("g", { transform: "translate(-7 0)" }, c);+ el("rect", { x: 131, y: -70, width: 8, height: 112, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, drv);+ el("rect", { x: 124, y: -80, width: 22, height: 13, rx: 3, fill: YEL, stroke: INK, "stroke-width": 2.5 }, drv);+ el("rect", { x: 115, y: -168, width: 40, height: 90, rx: 16, fill: LIL, stroke: INK, "stroke-width": 3 }, drv);+ const grip = el("path", { stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, drv);+ const turn = el("path", { d: "M100 -108 q-13 14 2 30 M170 -138 q13 14 -2 30", fill: "none", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, drv);+ star(svg, 606, 150, 15, YEL); star(svg, 886, 466, 17, MINT); star(svg, 598, 404, 12, SKY);+ return (t) => {+ bubble.setAttribute("cx", (-6 + 15 * Math.exp(-1.1 * t) * Math.cos(4.2 * t)).toFixed(2));+ const k = (t * 0.9) % 1;+ grip.setAttribute("d", [0, 1, 2].map((j) => `M${(123 + ((j + k) % 3) * 8).toFixed(1)} -152 V-96`).join(" "));+ turn.setAttribute("transform", `translate(0 ${(3 * Math.sin(t * 5)).toFixed(2)})`);+ };+ },+ };++ /* ---- office: a job slip lands on the front desk, is torn in three, and one piece comes back as a decision ---- */+ SC.office = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "What is Munder Difflin?", "tap the front desk");+ const BX = [118, 306, 494, 682], COL = [SKY, LIL, MINT, BLUE], SY = 172, FX = 400, FY = 354;+ const pp = (v, L) => L - Math.abs((((v % (2 * L)) + 2 * L) % (2 * L)) - L);+ el("path", { d: "M60 130 V108 H306 M494 108 H740 V130", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round", "stroke-dasharray": "1 9" }, svg);+ txt(svg, 400, 116, "other agents", { "text-anchor": "middle", "font-size": 24, "font-weight": 700, fill: MUTE });+ el("path", { d: "M44 492 H756", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ /* a filing cabinet and a water cooler, because it is an office */+ el("rect", { x: 62, y: 376, width: 108, height: 116, rx: 6, fill: LIL, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M62 434 H170", stroke: INK, "stroke-width": 3 }, svg);+ [398, 456].forEach((y) => el("rect", { x: 98, y, width: 36, height: 10, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 662, y: 318, width: 52, height: 86, rx: 16, fill: SKY, stroke: INK, "stroke-width": 3 }, svg);+ const bub = [0, 1].map(() => el("circle", { fill: "#fff", stroke: INK, "stroke-width": 2 }, svg));+ el("rect", { x: 652, y: 398, width: 72, height: 94, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 678, y: 420, width: 20, height: 12, rx: 3, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M670 452 H706", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, svg);+ /* the back row: four desks on a step, each with a screen */+ el("rect", { x: 30, y: 270, width: 740, height: 20, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ const desks = BX.map((x, i) => {+ el("rect", { x: x - 62, y: 228, width: 124, height: 43, fill: COL[i], stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: x - 74, y: 216, width: 148, height: 13, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: x - 46, y: 243, width: 92, height: 14, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const bar = el("rect", { x: x - 43, y: 246, width: 0, height: 8, rx: 4, fill: MINT }, svg);+ const mon = el("g", {}, svg);+ el("rect", { x: -7, y: -2, width: 14, height: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, mon);+ el("rect", { x: -50, y: -72, width: 100, height: 72, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, mon);+ /* the desk with no work has found the screensaver */+ const saver = i === 3 ? el("rect", { width: 18, height: 18, rx: 5, fill: LIL, stroke: INK, "stroke-width": 2.5 }, mon) : null;+ el("rect", { x: x - 26, y: 210, width: 52, height: 7, rx: 3, fill: INK }, svg);+ const tick = el("g", {}, svg);+ el("circle", { r: 13, fill: MINT, stroke: INK, "stroke-width": 2.5 }, tick);+ el("path", { d: "M-6 0 l4.5 5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, tick);+ return { x, bar, mon, saver, tick };+ });+ /* the front desk */+ const front = el("g", {}, svg);+ el("rect", { x: 192, y: 296, width: 416, height: 204, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, front);+ el("rect", { x: 220, y: 434, width: 360, height: 58, fill: YEL, stroke: INK, "stroke-width": 3 }, front);+ el("rect", { x: 200, y: 420, width: 400, height: 15, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, front);+ txt(front, 400, 473, "your orchestrator", { "text-anchor": "middle", "font-size": 27, "font-weight": 700 });+ const fmon = el("g", {}, front);+ el("rect", { x: -9, y: -2, width: 18, height: 14, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, fmon);+ el("rect", { x: -100, y: -100, width: 200, height: 100, rx: 9, fill: "#fff", stroke: INK, "stroke-width": 3 }, fmon);+ el("rect", { x: 364, y: 414, width: 72, height: 7, rx: 3, fill: INK }, front);+ button(front, "Send another job to the front desk", () => api.seek(0.3));+ /* the slips */+ const J = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -62, y: -36, width: 124, height: 72, rx: 7, fill: SOFT, stroke: INK, "stroke-width": 3 }, J);+ const jt = txt(J, 0, 9, "your job", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 });+ const perf = el("path", { d: "M-21 -30 V30 M21 -30 V30", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "2 7" }, J);+ const S = [SKY, LIL, MINT].map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ const r = el("rect", { x: -20, y: -36, width: 40, height: 72, rx: 6, fill: c, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-10 -16 H10 M-10 -2 H10 M-10 12 H4", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, r, c };+ });+ const D = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -80, y: -40, width: 160, height: 80, rx: 8, fill: YEL, stroke: INK, "stroke-width": 3 }, D);+ txt(D, 0, -5, "needs your", { "text-anchor": "middle", "font-size": 24, "font-weight": 700 });+ txt(D, 0, 24, "decision", { "text-anchor": "middle", "font-size": 24, "font-weight": 700 });+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 14, "#fff");+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out;+ bub.forEach((c, j) => {+ const p = (now * 0.3 + j / 2) % 1;+ c.setAttribute("cx", (688 + 8 * Math.sin(p * 6 + j * 2.4)).toFixed(1)); c.setAttribute("cy", (392 - 56 * p).toFixed(1));+ c.setAttribute("r", (2.5 + 3.5 * p).toFixed(1)); c.setAttribute("opacity", (cl(p * 6) * (1 - p * p)).toFixed(2));+ });+ /* the job arrives and is torn in three */+ const h = io(seg(t, 1.2, 2.1)), shake = t > 2.7 && t < 3 ? 3 * Math.sin(t * 70) : 0;+ tf(J, (lerp(116, FX, h) + shake).toFixed(2), (lerp(338, FY, h) - 40 * Math.sin(Math.PI * h)).toFixed(2), land(seg(t, 0.25, 0.65), 2).toFixed(3), ((1 - h) * -4 + 10 * Math.sin(Math.PI * h)).toFixed(2));+ J.setAttribute("opacity", (1 - seg(t, 3, 3.1)).toFixed(2));+ jt.setAttribute("opacity", (1 - seg(t, 2.25, 2.45)).toFixed(2));+ perf.setAttribute("opacity", seg(t, 2.45, 2.65).toFixed(2));+ S.forEach((s, i) => {+ const a = 3.6 + 0.4 * i, k = io(seg(t, a, a + 0.9)), fan = oc(seg(t, 3, 3.4)), stuck = i === 1 && t > 6.6;+ tf(s.g, lerp(FX + (i - 1) * lerp(41, 58, fan), BX[i], k).toFixed(2), (lerp(FY, SY, k) - 34 * Math.sin(Math.PI * k)).toFixed(2), lerp(1, 0.8, k).toFixed(3), ((i - 1) * 7 * fan * (1 - k) + (stuck ? 5 * Math.sin(t * 13) : 0)).toFixed(2));+ s.r.setAttribute("fill", stuck ? YEL : s.c);+ s.g.setAttribute("opacity", t < 3 ? 0 : i === 1 ? (1 - seg(t, 8.8, 8.95)).toFixed(2) : vis.toFixed(2));+ });+ desks.forEach((d, i) => {+ const tau = t - (4.5 + 0.4 * i);+ d.mon.setAttribute("transform", `translate(${d.x} 208) rotate(${(i < 3 && tau > 0 ? 4 * Math.exp(-6 * tau) * Math.sin(18 * tau) : 0).toFixed(2)})`);+ const p = i === 0 ? io(seg(t, 4.7, 7)) : i === 1 ? 0.55 * oc(seg(t, 5.1, 6.4)) : i === 2 ? io(seg(t, 5.5, 8.2)) : 0;+ d.bar.setAttribute("width", (86 * p * vis).toFixed(1));+ d.bar.setAttribute("fill", i === 1 && t > 6.6 ? YEL : MINT);+ const T = i === 0 ? 7 : i === 2 ? 8.2 : 99;+ tf(d.tick, d.x + 50, 136, (land(seg(t, T, T + 0.4), 2.4) * vis).toFixed(3));+ if (d.saver) { d.saver.setAttribute("x", (-41 + pp(now * 30 + 22, 64)).toFixed(1)); d.saver.setAttribute("y", (-64 + pp(now * 23 + 9, 38)).toFixed(1)); }+ });+ /* one piece comes back */+ const e = io(seg(t, 8.8, 9.8)), tau = t - 9.8, w1 = t - 2.1, wob = (v) => (v > 0 ? Math.exp(-6 * v) * Math.sin(17 * v) : 0);+ tf(D, lerp(BX[1], FX, e).toFixed(2), (lerp(SY, FY, e) - 30 * Math.sin(Math.PI * e)).toFixed(2), (lerp(0.22, 1, e) * (1 + (tau > 0 ? 0.1 * Math.exp(-5 * tau) * Math.sin(15 * tau) : 0) + 0.02 * seg(t, 10.6, 11.4) * Math.sin(now * 3))).toFixed(4), ((1 - e) * -12 + 3 * wob(tau)).toFixed(2));+ D.setAttribute("opacity", t >= 8.8 ? vis.toFixed(2) : 0);+ fmon.setAttribute("transform", `translate(400 404) rotate(${(2.2 * (wob(w1) + wob(tau))).toFixed(2)})`);+ tf(sp, 532, 316, (land(seg(t, 9.8, 10.3), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.2 || out > 0 ? "A small office of AI workers" : t < 2.6 ? "You brief one orchestrator" : t < 3.6 ? "He splits it into pieces" : t < 8.8 ? "He hands the pieces to other agents" : "Only the decisions come back to you", 700);+ };+ },+ };++ /* ---- rooms: a house plan where every room is a folder; a new room is built and the agent's desk goes only there ---- */+ SC.rooms = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "What can it touch on your computer?", "tap a room");+ const R = [+ { n: "Documents", x: 44, y: 96, w: 236, h: 180, c: SKY, dx: 226, dy: 226 },+ { n: "Photos", x: 280, y: 96, w: 238, h: 180, c: YEL, dx: 466, dy: 226 },+ { n: "Projects", x: 518, y: 96, w: 238, h: 180, c: SKY, dx: 704, dy: 226 },+ { n: "Downloads", x: 44, y: 276, w: 256, h: 214, c: YEL, dx: 246, dy: 408 },+ ];+ const NX = 300, NY = 276, NW = 456, NH = 214, FOLD = "M0 -11 H11 L16 -6 H30 V11 H0 Z", NEW = "M300 490 H756 V276";+ const given = [false, false, false, false], gt = [-99, -99, -99, -99];+ R.forEach((r) => {+ el("rect", { x: r.x, y: r.y, width: r.w, height: r.h, fill: SOFT }, svg);+ r.fl = el("rect", { x: r.x, y: r.y, width: r.w, height: r.h, fill: r.c, opacity: 0 }, svg);+ r.in = el("g", {}, svg);+ });+ /* the plot where the new room will go: hatched, until it has a floor */+ const hatch = el("path", { d: [0, 1, 2, 3, 4, 5, 6].map((k) => `M${318 + 48 * k} 480 l136 -136`).join(" "), stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "6 9" }, svg);+ const nfl = el("rect", { x: NX, y: NY, width: NW, height: NH, fill: MINT, opacity: 0 }, svg);+ const sheet = (p, x, y, r) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, p);+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-11 -12 H11 M-11 -2 H11 M-11 8 H5", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ return g;+ };+ /* Documents: two tidy stacks. Photos: two prints. Projects: two folders. */+ [88, 146].forEach((x) => [8, 4, 0].forEach((o) => sheet(R[0].in, x + o, 224 - o, 0)));+ [[330, 224, -7], [390, 218, 5]].forEach(([x, y, r]) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, R[1].in);+ el("rect", { x: -27, y: -23, width: 54, height: 46, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -20, y: -16, width: 40, height: 26, fill: SKY, stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-20 10 L-7 -5 L2 4 L8 -2 L20 10 Z", fill: MINT, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, g);+ el("circle", { cx: 11, cy: -8, r: 4, fill: YEL, stroke: INK, "stroke-width": 2 }, g);+ });+ [[540, LIL], [598, MINT]].forEach(([x, c]) => el("path", { d: "M0 -18 H16 L23 -11 H46 V20 H0 Z", fill: c, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} 222)` }, R[2].in));+ /* Downloads: a heap, and it keeps growing */+ [[-32, -18, -22], [26, -22, 14], [-8, 4, 38], [32, 16, -9], [-34, 22, 8], [4, -30, -31], [-18, 28, -47], [30, -4, 62], [0, 6, -4]].forEach(([dx, dy, r]) => sheet(R[3].in, 122 + dx, 410 + dy, r));+ const drops = [[14, -14, 21, 3.2], [-24, 2, -15, 7], [20, 22, 33, 10.4]].map(([dx, dy, r, at]) => ({ g: sheet(R[3].in, 0, 0, 0), x: 122 + dx, y: 410 + dy, r, at }));+ /* walls: the plot for the new room is dotted until it is built */+ el("path", { d: NEW, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg);+ const nw = el("path", { d: NEW, fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round", "stroke-dasharray": "670 670" }, svg);+ el("path", { d: "M44 96 H756 V276 H44 Z M280 96 V276 M518 96 V276 M44 276 V490 H300 V276", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ const door = (p, gap, leaf) => { el("path", { d: gap, stroke: "#fff", "stroke-width": 8 }, p); el("path", { d: leaf, fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, p); };+ door(svg, "M280 156 V192", "M280 192 H244 A36 36 0 0 1 280 156");+ door(svg, "M518 156 V192", "M518 192 H482 A36 36 0 0 1 518 156");+ const nd = el("g", {}, svg); door(nd, "M300 438 V478", "M300 478 H340 A40 40 0 0 0 300 438");+ /* each room is labelled as a folder */+ R.forEach((r) => {+ r.ic = el("path", { d: FOLD, fill: "#fff", stroke: MUTE, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${r.x + 18} ${r.y + 34})` }, svg);+ r.tx = txt(svg, r.x + 56, r.y + 43, r.n, { "font-size": 25, "font-weight": 700, fill: MUTE });+ });+ const plus = el("g", {}, svg);+ el("circle", { r: 36, fill: "#fff", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, plus);+ el("path", { d: "M-15 0 H15 M0 -15 V15", stroke: LIL, "stroke-width": 6, "stroke-linecap": "round" }, plus);+ const nl = el("g", {}, svg);+ const ni = el("path", { d: FOLD, fill: "#fff", stroke: MUTE, "stroke-width": 2.5, "stroke-linejoin": "round", transform: "translate(0 -9)" }, nl);+ const nt = txt(nl, 40, 0, "new empty folder", { "font-size": 26, "font-weight": 700, fill: MUTE });+ const files = [[356, 6, -5], [414, 7.2, 3], [472, 8.4, -2]].map(([x, at, r]) => ({ g: mv(sheet(svg, 0, 0, 0)), x, at, r }));+ /* a desk seen from above, with a small terminal on it */+ const desk = () => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -14, y: 14, width: 28, height: 24, rx: 9, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -42, y: -24, width: 84, height: 46, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: -22, y: -16, width: 44, height: 30, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -22, y: -16, width: 44, height: 9, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-14 -2 l5 4 l-5 4", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return { g, cur: el("path", { d: "M-3 6 H8", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g) };+ };+ R.forEach((r) => (r.desk = desk()));+ const ad = desk(), tag = mv(pill(svg, "agent works here", YEL));+ tag.setAttribute("pointer-events", "none");+ R.forEach((r, i) => {+ const g = el("g", {}, svg);+ el("rect", { x: r.x + 9, y: r.y + 9, width: r.w - 18, height: r.h - 18, rx: 10, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ button(g, `Give the agent the ${r.n} folder`, () => { given[i] = !given[i]; gt[i] = api.now(); api.poke(); });+ });+ const ng = el("g", {}, svg);+ el("rect", { x: NX + 9, y: NY + 9, width: NW - 18, height: NH - 18, rx: 10, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, ng);+ button(ng, "Make the new empty folder again", () => api.seek(1.2));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, blink = Math.sin(now * 6) > 0 ? 1 : 0;+ /* the new room is built */+ const built = io(seg(t, 1.6, 2.5)) * (1 - io(out));+ nw.setAttribute("stroke-dashoffset", (670 * (1 - built)).toFixed(1));+ nw.setAttribute("opacity", built > 0.002 ? 1 : 0);+ nfl.setAttribute("opacity", (seg(t, 2.2, 2.9) * vis).toFixed(2));+ hatch.setAttribute("opacity", (1 - seg(t, 1.9, 2.3) * (1 - seg(t, 13.5, 13.8))).toFixed(2));+ nd.setAttribute("opacity", (seg(t, 2.6, 2.9) * vis).toFixed(2));+ const made = t >= 2.4 && t < 13.4;+ tf(nl, 320, 320, (1 + 0.1 * Math.sin(Math.PI * seg(t, 2.4, 2.85))).toFixed(3));+ nt.setAttribute("fill", made ? INK : MUTE); ni.setAttribute("stroke", made ? INK : MUTE);+ tf(plus, 528, 412, ((1 - oc(seg(t, 1.5, 1.9)) + oc(seg(t, 13.4, 13.8))) * (1 + 0.05 * Math.sin(now * 3))).toFixed(3));+ /* the desk moves in, and work starts */+ const dk = seg(t, 3.6, 4.2);+ tf(ad.g, 630, (380 - 30 * (1 - oc(dk))).toFixed(2), (lerp(1.5, 1, cl(bk(dk, 2), 0, 1.3)) * (dk > 0 ? vis : 0)).toFixed(3), ((1 - oc(dk)) * -10).toFixed(2));+ ad.cur.setAttribute("opacity", blink);+ const tk = seg(t, 4.5, 4.95);+ tf(tag, 630, 448, (land(tk) * vis).toFixed(3), ((1 - oc(tk)) * -8).toFixed(2));+ files.forEach((f) => tf(f.g, f.x, 400, (land(seg(t, f.at, f.at + 0.4), 2.4) * vis).toFixed(3), f.r));+ drops.forEach((d) => {+ const k = seg(t, d.at, d.at + 0.45), w = t - d.at - 0.45;+ tf(d.g, d.x, d.y, (k > 0 ? lerp(1.7, 1, cl(bk(k, 1.6), 0, 1.2)) * vis : 0).toFixed(3), (d.r + (1 - oc(k)) * 40 + (w > 0 ? 6 * Math.exp(-5 * w) * Math.sin(15 * w) : 0)).toFixed(2));+ });+ /* any room you tap is given to the agent as well */+ let fresh = false;+ R.forEach((r, i) => {+ const w = now - gt[i], g = given[i] ? oc(seg(w, 0, 0.45)) : 1 - oc(seg(w, 0, 0.35));+ r.fl.setAttribute("opacity", g.toFixed(2));+ r.in.setAttribute("opacity", lerp(0.42, 1, g).toFixed(2));+ r.tx.setAttribute("fill", g > 0.5 ? INK : MUTE); r.ic.setAttribute("stroke", g > 0.5 ? INK : MUTE);+ tf(r.desk.g, r.dx, r.dy, (given[i] ? land(seg(w, 0.12, 0.6), 2.2) : 1 - seg(w, 0, 0.2)).toFixed(3), (given[i] ? (1 - oc(seg(w, 0.12, 0.6))) * -10 : 0).toFixed(2));+ r.desk.cur.setAttribute("opacity", blink);+ if (given[i] && w < 4.5) fresh = true;+ });+ say(cap, fresh ? "Agents work inside the folders you give them" : t < 1.6 || out > 0 ? "You choose the folders" : t < 3.6 ? "For your first job, make a new empty folder" : "An agent works only in the folders you give it", 700);+ };+ },+ };++ /* ---- engine: three cards on a table, one per thing you might pay for; each turns over to show its engine ---- */+ SC.engine = {+ dur: 14, still: 11.5,+ build(svg, api) {+ head(svg, "Which AI engine should you pick?", "tap a card");+ const X = [150, 400, 650], CY = 284, COL = [YEL, SKY, MINT], REST = [-2.5, 1.5, -1.5], F = [2, 4, 6], HOLD = 4.5;+ const PAY = ["Claude", "ChatGPT", "nothing"], WITH = [["with a Claude", "subscription"], ["with a ChatGPT", "subscription"], ["if you pay", "for nothing"]];+ const mid = (a) => Object.assign({ "text-anchor": "middle" }, a);+ const mats = X.map(() => el("rect", { x: -114, y: -176, width: 228, height: 352, rx: 24, fill: SOFT }, svg));+ const cards = X.map((x, i) => {+ const g = mv(el("g", {}, svg));+ el("rect", { x: -111, y: -173, width: 222, height: 346, rx: 22, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const fr = el("g", {}, g), ba = el("g", {}, g);+ /* the side that asks */+ el("rect", { x: -106, y: -168, width: 212, height: 336, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 3 }, fr);+ el("circle", { cx: -78, cy: -140, r: 11, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, fr);+ el("circle", { cx: 78, cy: 140, r: 11, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, fr);+ txt(fr, 0, -18, "You pay for", mid({ "font-size": 25, "font-weight": 500 }));+ const pw = fit(txt(fr, 0, 34, PAY[i], mid({ "font-size": 46, "font-weight": 700 })), 184).getComputedTextLength();+ el("path", { d: `M${(-pw / 2 + 4).toFixed(1)} 58 H${(pw / 2 - 4).toFixed(1)}`, stroke: COL[i], "stroke-width": 10, "stroke-linecap": "round" }, fr);+ /* the side that answers */+ el("rect", { x: -106, y: -168, width: 212, height: 336, rx: 18, fill: COL[i], stroke: INK, "stroke-width": 3 }, ba);+ el("circle", { cx: -78, cy: -140, r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ba);+ el("circle", { cx: 78, cy: 140, r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ba);+ txt(ba, 0, -64, "use", mid({ "font-size": 22, "font-weight": 500 }));+ fit(txt(ba, 0, -22, ["Claude Code", "Codex", "Antigravity"][i], mid({ "font-size": 32, "font-weight": 700 })), 188);+ el("path", { d: "M-74 8 H74", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, ba);+ WITH[i].forEach((s, j) => txt(ba, 0, 56 + j * 30, s, mid({ "font-size": 22, "font-weight": 600 })));+ return { g, fr, ba };+ });+ const st = X.map(() => ({ a: 0, m: null }));+ cards.forEach((c, i) => button(c.g, `Turn over the card for ${PAY[i]}`, () => {+ const back = Math.cos(Math.PI * st[i].a) < 0;+ st[i].m = { at: api.now(), a0: back ? 1 : 0, a1: back ? 0 : 1 };+ api.poke();+ }));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const end = seg(t, 12.9, 13.1);+ cards.forEach((c, i) => {+ const s = st[i], kin = seg(t, 0.1 + 0.18 * i, 0.6 + 0.18 * i), ku = seg(t, F[i], F[i] + (i === 2 ? 1.5 : 0.8)), kd = seg(t, 13.1 + 0.1 * i, 13.6 + 0.1 * i);+ /* the card that costs nothing turns over once more than it needs to, and lands crooked */+ let a = (i === 2 ? 3 : 1) * io(ku) - io(kd), lift = (i === 2 ? 1.4 : 1) * Math.sin(Math.PI * ku) + Math.sin(Math.PI * kd), crook = i === 2 ? 5 * io(seg(t, 6.7, 7.5)) * (1 - cl(bk(seg(t, 8.1, 8.55), 1.8), 0, 1.2)) : 0;+ if (s.m) {+ const w = now - s.m.at, goal = Math.cos(Math.PI * a) < 0 ? 1 : 0;+ if (w < HOLD) { const k = seg(w, 0, 0.6); a = lerp(s.m.a0, s.m.a1, io(k)); lift = Math.sin(Math.PI * k); crook = 0; }+ else if (goal === s.m.a1 || w > HOLD + 0.6) s.m = null;+ else { const k = seg(w - HOLD, 0, 0.6); a = lerp(s.m.a1, goal, io(k)); lift = Math.sin(Math.PI * k); crook = 0; }+ }+ s.a = a;+ const cs = Math.cos(Math.PI * a), back = cs < 0, sx = Math.max(Math.abs(cs), 0.02), sc = 1 + 0.07 * lift;+ const rot = REST[i] + (1 - oc(kin)) * (i % 2 ? 3 : -3) + crook + 0.6 * Math.sin(now * 1.1 + i * 2.1) * seg(t, 8.8, 9.6) * (1 - end);+ const y = CY + 26 * (1 - oc(kin)) - 16 * lift;+ c.g.setAttribute("transform", `translate(${X[i]} ${y.toFixed(2)}) rotate(${rot.toFixed(2)}) scale(${(sx * sc).toFixed(4)} ${sc.toFixed(4)})`);+ c.fr.setAttribute("opacity", back ? 0 : 1); c.ba.setAttribute("opacity", back ? 1 : 0);+ mats[i].setAttribute("transform", `translate(${X[i]} ${(CY + 26 * (1 - oc(kin))).toFixed(2)}) rotate(${REST[i]}) scale(${(1 + 0.03 * lift).toFixed(4)})`);+ });+ say(cap, t < 2 || t > 13.1 ? "What you already pay for usually decides it" : t < 4 ? "You pay for Claude: use Claude Code" : t < 6 ? "You pay for ChatGPT: use Codex" : t < 8.7 ? "You pay for nothing: use Antigravity" : "You need exactly one to start", 700);+ };+ },+ };++ /* ---- terminal: a light text window above a row of keys; three lines are typed and each one is sent with Enter ---- */+ SC.terminal = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Install an engine and sign in", "tap the terminal");+ const CMD = ["npm install -g @openai/codex", "codex", "codex --version"];+ const X0 = 92, BY = [160, 228, 296, 362], TY = [0.7, 5.7, 9.5], TD = [2, 0.5, 1.1], ENT = [2.9, 6.4, 10.8], PR = [0, 5.45, 9.25, 11.7];+ const replay = () => api.seek(0.3);+ const tickDot = (p, r) => {+ const g = el("g", {}, p);+ el("circle", { r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${(-0.46 * r).toFixed(1)} ${(0.04 * r).toFixed(1)} l${(0.34 * r).toFixed(1)} ${(0.36 * r).toFixed(1)} l${(0.6 * r).toFixed(1)} ${(-0.76 * r).toFixed(1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* the keys: seven plain ones and Enter */+ const keys = [0, 1, 2, 3, 4, 5, 6].map((k) => el("rect", { x: 44 + 78 * k, y: 404, width: 68, height: 64, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg));+ const ek = el("g", {}, svg);+ el("rect", { x: 590, y: 404, width: 166, height: 64, rx: 12, fill: YEL, stroke: INK, "stroke-width": 3 }, ek);+ txt(ek, 652, 445, "Enter", { "text-anchor": "middle", "font-size": 26, "font-weight": 700 });+ el("path", { d: "M728 424 V438 H702 M710 430 l-8 8 l8 8", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, ek);+ ek.classList.add("mg-hit"); ek.addEventListener("click", replay);+ /* the window: a white title bar and a soft page, ink type */+ const term = el("g", {}, svg);+ el("rect", { x: 37, y: 85, width: 726, height: 306, rx: 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, term);+ el("rect", { x: 44, y: 92, width: 712, height: 292, rx: 14, fill: "#fff" }, term);+ el("path", { d: "M44 130 H756 V370 a14 14 0 0 1 -14 14 H58 a14 14 0 0 1 -14 -14 Z", fill: SOFT }, term);+ el("path", { d: "M44 130 H756", stroke: INK, "stroke-width": 3 }, term);+ el("rect", { x: 44, y: 92, width: 712, height: 292, rx: 14, fill: "none", stroke: INK, "stroke-width": 3 }, term);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 20 * i, cy: 111, r: 6, fill: c, stroke: INK, "stroke-width": 2.5 }, term));+ txt(term, 400, 118, "Terminal", { "text-anchor": "middle", "font-size": 20, fill: MUTE });+ const chev = BY.map((y) => el("path", { d: `M70 ${y - 14} l8 7 l-8 7`, fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, term));+ const body = el("g", {}, term);+ const line = CMD.map((s, i) => { const n = txt(body, X0, BY[i], s, { "font-family": MONO, "font-size": 22, "font-weight": 700 }); n.style.whiteSpace = "pre"; return n; });+ const CW = line[0].getComputedTextLength() / CMD[0].length;+ const cur = el("rect", { y: -19, width: 12, height: 24, rx: 2, fill: INK }, term);+ /* what each line prints: a progress bar, a wait for the browser, a version */+ const bar = el("g", {}, body);+ const blk = Array.from({ length: 12 }, (_, j) => el("rect", { x: X0 + 29 * j, y: 176, width: 24, height: 16, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, bar));+ const wait = [0, 1, 2].map((j) => el("circle", { cx: 98 + 16 * j, cy: 252, r: 4.5, fill: INK }, body));+ const arrow = el("path", { fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, body);+ const ahead = el("path", { d: "M-9 -8 L0 0 L-9 8", fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, body);+ const ver = [0, 1, 2].map(() => el("rect", { x: -7.5, y: -9, width: 15, height: 18, rx: 4, fill: INK }, body));+ const vdot = [0, 1].map((j) => el("circle", { cx: X0 + 20 + 25 * j, cy: 327, r: 2.3, fill: INK }, body));+ const vt = txt(body, 202, 327, "a version number", { "font-size": 20 });+ const tk = [0, 1, 2].map(() => tickDot(body, 11));+ /* the browser that the second line opens */+ const B = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -108, y: -98, width: 216, height: 196, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, B);+ el("path", { d: "M-108 -64 H108", stroke: INK, "stroke-width": 3 }, B);+ [0, 1, 2].forEach((j) => el("circle", { cx: -90 + 16 * j, cy: -81, r: 5, fill: SOFT, stroke: INK, "stroke-width": 2 }, B));+ el("rect", { x: -36, y: -91, width: 128, height: 20, rx: 10, fill: SOFT }, B);+ txt(B, 0, -16, "Sign in", { "text-anchor": "middle", "font-size": 28, "font-weight": 700 });+ const sb = el("g", {}, B);+ el("rect", { x: -62, y: -20, width: 124, height: 40, rx: 20, fill: YEL, stroke: INK, "stroke-width": 3 }, sb);+ el("path", { d: "M-16 0 H14 M6 -8 l8 8 l-8 8", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, sb);+ const bt = tickDot(B, 24);+ button(term, "Run the three terminal steps again", replay);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, on = Math.sin(now * 6) >= 0 ? 1 : 0;+ body.setAttribute("opacity", vis.toFixed(2));+ /* typing: one key goes down for each letter, then Enter */+ let row = -1, col = 0, typing = false, key = -1, ph = 0;+ CMD.forEach((s, i) => {+ const k = seg(t, TY[i], TY[i] + TD[i]), n = Math.min(s.length, Math.floor(k * s.length + 1e-6)), shown = s.slice(0, n);+ if (line[i]._s !== shown) { line[i]._s = shown; line[i].textContent = shown; }+ if (t >= PR[i] && t < ENT[i] + 0.1) { row = i; col = n; if (k > 0 && k < 1) { typing = true; key = (n * 3 + i * 2 + 1) % 7; ph = k * s.length - n; } }+ });+ if (t >= PR[3] && t < 13.2) { row = 3; col = 0; }+ if (t >= 13.7) { row = 0; col = 0; }+ chev.forEach((c, i) => c.setAttribute("opacity", t >= PR[i] ? (i ? vis.toFixed(2) : 1) : 0));+ cur.setAttribute("transform", `translate(${(X0 + col * CW + (col ? 3 : 0)).toFixed(1)} ${BY[Math.max(row, 0)]})`);+ cur.setAttribute("opacity", row < 0 ? 0 : typing ? 1 : on);+ keys.forEach((r, k) => { const p = k === key ? Math.sin(Math.PI * ph) : 0; r.setAttribute("y", (404 + 5 * p).toFixed(2)); r.setAttribute("fill", p > 0.25 ? LIL : "#fff"); });+ const ep = Math.max(...ENT.map((e) => Math.sin(Math.PI * seg(t, e - 0.04, e + 0.26))));+ ek.setAttribute("transform", `translate(${(673 * 0.04 * ep).toFixed(2)} ${(6 * ep + 436 * 0.04 * ep).toFixed(2)}) scale(${(1 - 0.04 * ep).toFixed(4)})`);+ /* line one: the bar rushes to the last block, loses one, then finishes */+ const nb = t < 3.1 ? 0 : t < 3.85 ? Math.floor(11 * oc(seg(t, 3.1, 3.8)) + 0.001) : t < 4.45 ? 11 : t < 5.05 ? 10 : 12;+ bar.setAttribute("opacity", t >= ENT[0] + 0.15 ? 1 : 0);+ bar.setAttribute("transform", `translate(${(t > 4.45 && t < 4.8 ? 2.5 * Math.sin((t - 4.45) * 55) * (1 - seg(t, 4.45, 4.8)) : 0).toFixed(2)} 0)`);+ blk.forEach((r, j) => r.setAttribute("fill", j < nb ? MINT : "#fff"));+ tf(tk[0], 452, 184, land(seg(t, 5.1, 5.5), 2.4).toFixed(3));+ /* line two: the browser slides in, you sign in there */+ const ak = io(seg(t, 6.55, 7)), ax = lerp(180, 520, ak), waiting = t >= ENT[1] + 0.15 && t < 8.7;+ arrow.setAttribute("d", `M180 221 H${ax.toFixed(1)}`); arrow.setAttribute("opacity", ak > 0.02 ? 1 : 0);+ ahead.setAttribute("transform", `translate(${(ax + 3).toFixed(1)} 221)`); ahead.setAttribute("opacity", ak > 0.06 ? 1 : 0);+ wait.forEach((c, j) => c.setAttribute("opacity", waiting ? (0.25 + 0.75 * Math.max(0, Math.sin(now * 5 - j * 0.9))).toFixed(2) : 0));+ tf(tk[1], 103, 252, land(seg(t, 8.7, 9.1), 2.4).toFixed(3));+ const bi = seg(t, 6.95, 7.6);+ tf(B, (640 + 70 * (1 - cl(bk(bi, 1.3), 0, 1.2)) + 70 * io(out)).toFixed(2), 262, lerp(0.9, 1, oc(bi)).toFixed(3), ((1 - oc(bi)) * 4).toFixed(2));+ B.setAttribute("opacity", (seg(bi, 0, 0.35) * (1 - seg(out, 0.3, 0.9))).toFixed(2));+ tf(sb, 0, 42, ((1 - 0.08 * Math.sin(Math.PI * seg(t, 8.05, 8.35))) * (1 - seg(t, 8.4, 8.55))).toFixed(3));+ tf(bt, 0, 42, (land(seg(t, 8.5, 8.9), 2.6) * (1 + 0.04 * seg(t, 9.6, 10.2) * Math.sin(now * 3))).toFixed(3));+ /* line three: a version number, drawn as blocks */+ ver.forEach((r, j) => tf(r, X0 + 7.5 + 25 * j, 320, land(seg(t, 11 + 0.1 * j, 11.25 + 0.1 * j), 2.2).toFixed(3)));+ vdot.forEach((c, j) => c.setAttribute("opacity", t >= 11.12 + 0.1 * j ? 1 : 0));+ tf(tk[2], 181, 320, land(seg(t, 11.35, 11.75), 2.4).toFixed(3));+ vt.setAttribute("opacity", seg(t, 11.4, 11.6).toFixed(2));+ say(cap, t < TY[0] || out > 0 ? "Open a terminal, paste one block, press Enter" : t < PR[1] ? "Install the engine" : t < PR[2] ? "Run it once to sign in" : t < 11.3 ? "Then ask for its version to check" : "A version number means you are done with the terminal", 700);+ };+ },+ };++ /* ---- download: the download page has one button per system; each lets go of its file, which comes down on a parachute ---- */+ SC.download = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Get the file for your system", "tap a button");+ const SYS = ["macOS", "Windows", "Linux"], COL = [SKY, LIL, MINT], BX = [132, 394, 668], FX = [132, 322, 668], START = [1, 4.4, 8.6], LEN = [2.7, 3.6, 2.7];+ const LINE = ["A universal .dmg for macOS", "A setup .exe or a portable .exe for Windows", "An .AppImage for Linux"];+ const YH = 248, YF = 346, YP = 374, YR = 416, PH = 256, PR = 422, tap = [null, null, null], mid = { "text-anchor": "middle" };+ const bo = (x) => { const n = 7.5625, d = 2.75; return x < 1 / d ? n * x * x : x < 2 / d ? n * (x -= 1.5 / d) * x + 0.75 : x < 2.5 / d ? n * (x -= 2.25 / d) * x + 0.9375 : n * (x -= 2.625 / d) * x + 0.984375; };+ /* the ground, with a landing pad for each system */+ el("path", { d: "M44 490 H756", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ [[46, 172], [238, 312], [582, 172]].forEach(([x, w], i) => el("rect", { x, y: 472, width: w, height: 18, rx: 6, fill: COL[i], stroke: INK, "stroke-width": 3 }, svg));+ const file = (p, w, h, c, a, b) => {+ const g = el("g", {}, p), f = 18;+ el("path", { d: `M${-w} ${-h} H${w - f} L${w} ${-h + f} V${h} H${-w} Z`, fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, g);+ el("path", { d: `M${w - f} ${-h} V${-h + f} H${w} Z`, fill: c, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, g);+ if (a) txt(g, 0, -h + 42, a, Object.assign({ "font-size": 19 }, mid));+ fit(txt(g, 0, a ? -h + 74 : -h + 63, b, Object.assign({ "font-family": MONO, "font-size": a ? 30 : 23, "font-weight": 700 }, mid)), 2 * w - 22);+ el("path", { d: h > 52 ? `M${-w + 18} ${h - 20} H${w - 18} M${-w + 18} ${h - 9} H${w - 52}` : `M${-w + 18} ${h - 10} H${w - 18}`, stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ };+ const guide = [132, 322, 478, 668].map((x) => {+ const g = el("g", {}, svg);+ el("path", { d: `M${x} 322 V446`, stroke: LIL, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, g);+ el("path", { d: `M${x - 10} 446 l10 10 l10 -10`, fill: "none", stroke: LIL, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ });+ const rigs = [["universal", ".dmg"], ["setup", ".exe"], [null, ".AppImage"]].map(([a, b], i) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ const str = el("path", { fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ const can = el("g", {}, g);+ el("path", { d: "M-75 0 C-75 -36 -42 -54 0 -54 C42 -54 75 -36 75 0 Q56 -11 37.5 0 Q19 -11 0 0 Q-19 -11 -37.5 0 Q-56 -11 -75 0 Z", fill: COL[i], stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, can);+ el("path", { d: "M0 -54 Q-24 -32 -37.5 0 M0 -54 V0 M0 -54 Q24 -32 37.5 0", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, can);+ return { g, str, can, f: file(g, 75, 56, COL[i], a, b) };+ });+ /* the portable one has a carry handle, and no parachute */+ const port = mv(el("g", { "pointer-events": "none" }, svg)), HANDLE = "M-16 -50 V-58 a7 7 0 0 1 7 -7 H9 a7 7 0 0 1 7 7 V-50";+ el("path", { d: HANDLE, fill: "none", stroke: INK, "stroke-width": 10, "stroke-linecap": "round" }, port);+ el("path", { d: HANDLE, fill: "none", stroke: YEL, "stroke-width": 4.5, "stroke-linecap": "round" }, port);+ file(port, 59, 50, LIL, "portable", ".exe");+ /* the download page */+ el("rect", { x: 44, y: 92, width: 712, height: 94, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M44 124 H756", stroke: INK, "stroke-width": 3 }, svg);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 262, y: 96, width: 276, height: 24, rx: 12, fill: SOFT }, svg);+ txt(svg, 400, 114, "harnessmd.com/download", Object.assign({ "font-family": MONO, "font-size": 17, "font-weight": 700 }, mid));+ [132, 322, 478, 668].forEach((x) => el("rect", { x: x - 7, y: 182, width: 14, height: 20, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg));+ const btn = SYS.map((s, i) => {+ const g = el("g", {}, svg), x = BX[i];+ el("rect", { x: x - 81, y: 128, width: 162, height: 54, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const f = el("g", {}, g);+ el("rect", { x: x - 76, y: 133, width: 152, height: 42, rx: 12, fill: COL[i], stroke: INK, "stroke-width": 3 }, f);+ el("path", { d: `M${x - 57} 144 V158 M${x - 63} 153 l6 6 l6 -6 M${x - 65} 165 H${x - 49}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, f);+ txt(f, x + 13, 162, s, Object.assign({ "font-size": 23, "font-weight": 700 }, mid));+ button(g, `Drop the ${s} file again`, () => { tap[i] = api.now(); api.poke(); });+ return f;+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const wrap = t >= 13.2;+ if (wrap) tap.fill(null);+ let hot = -1;+ rigs.forEach((R, i) => {+ const lc = t - START[i];+ if (tap[i] !== null && now - tap[i] >= LEN[i] && lc >= LEN[i]) tap[i] = null;+ const c = tap[i] !== null ? now - tap[i] : wrap ? (t < 13.55 ? 99 : -1) : lc;+ if (tap[i] !== null && c < 3.6) hot = i;+ const a = (wrap ? (t < 13.55 ? 1 - seg(t, 13.2, 13.5) : seg(t, 13.6, 13.95)) : 1) * (tap[i] !== null ? seg(c, 0, 0.18) : 1);+ const hop = wrap && t >= 13.55 ? -8 * (1 - oc(seg(t, 13.6, 13.95))) : 0;+ /* let go, fall, open, float, land */+ let y = YH, s = 0, sw = 0, q = 0;+ if (c > 0.2 && c < 0.65) { const k = (c - 0.2) / 0.45; y = lerp(YH, YF, k * k); }+ else if (c >= 0.65) {+ s = land(seg(c, 0.65, 1), 2);+ y = c < 0.95 ? lerp(YF, YP, oc((c - 0.65) / 0.3)) : lerp(YP, YR, io(seg(c, 0.95, 2.2)));+ sw = Math.sin((c - 0.65) * 4.4) * (1 - seg(c, 1.4, 2.2));+ if (c > 2.2) q = Math.exp(-7 * (c - 2.2)) * Math.cos(18 * (c - 2.2));+ }+ const idle = seg(c, 2.4, 3.2), dx = 2.5 * idle * Math.sin(now * 1.3 + i * 2), sy = s * (1 + 0.04 * idle * Math.sin(now * 2.2 + i * 1.7));+ R.g.setAttribute("transform", `translate(${(FX[i] + 7 * sw).toFixed(2)} ${(y + hop).toFixed(2)}) rotate(${(5 * sw).toFixed(2)})`);+ R.g.setAttribute("opacity", a.toFixed(2));+ R.f.setAttribute("transform", `translate(0 56) scale(${(1 + 0.04 * q).toFixed(4)} ${(1 - 0.06 * q).toFixed(4)}) translate(0 -56)`);+ if (s < 0.02) { R.can.setAttribute("display", "none"); R.str.setAttribute("display", "none"); }+ else {+ R.can.removeAttribute("display"); R.str.removeAttribute("display");+ R.can.setAttribute("transform", `translate(${dx.toFixed(2)} -106) scale(${s.toFixed(4)} ${sy.toFixed(4)})`);+ R.str.setAttribute("d", [[-75, -46], [-37.5, -46], [37.5, 46], [75, 46]].map(([u, v]) => `M${(u * s + dx).toFixed(1)} -106 L${v} -56`).join(" "));+ }+ btn[i].setAttribute("transform", `translate(0 ${(4.5 * Math.sin(Math.PI * seg(c, 0, 0.3))).toFixed(2)})`);+ const gv = wrap ? (t < 13.55 ? 0 : seg(t, 13.6, 13.95)) : 1, fade = (j, cc) => guide[j].setAttribute("opacity", (gv * (1 - seg(cc, 0.05, 0.4))).toFixed(2));+ if (i === 1) { fade(1, c); fade(2, c - 1.3); } else fade(i ? 3 : 0, c);+ if (i === 1) {+ const k = seg(c - 1.3, 0, 1);+ tf(port, 478, (lerp(PH, PR, bo(k)) + hop).toFixed(2), 1, (k > 0 && k < 1 ? 9 * Math.sin(k * 10) * (1 - k) : 0).toFixed(2));+ port.setAttribute("opacity", a.toFixed(2));+ }+ });+ say(cap, hot >= 0 ? LINE[hot] : t < START[0] || wrap ? "Get the file for your system from the download page" : t < START[1] ? LINE[0] : t < START[2] ? LINE[1] : t < 11.4 ? LINE[2] : "One file per system", 700);+ };+ },+ };++ /* ---- signin: two windows; the browser shows a code and a copy of it is carried across to the app ---- */+ SC.signin = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Sign in to continue", "tap the app window");+ const AX = 222, FY = 306, mid = { "text-anchor": "middle" };+ const tickDot = (p, r) => {+ const g = el("g", {}, p);+ el("circle", { r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${(-0.46 * r).toFixed(1)} ${(0.04 * r).toFixed(1)} l${(0.34 * r).toFixed(1)} ${(0.36 * r).toFixed(1)} l${(0.6 * r).toFixed(1)} ${(-0.76 * r).toFixed(1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* the code is never letters: six boxes, a dot in each */+ const chip = (p) => {+ const g = el("g", {}, p), d = [];+ for (let j = 0; j < 6; j++) {+ const b = el("g", { transform: `translate(${-95 + 38 * j} 0)` }, g);+ el("rect", { x: -16, y: -20, width: 32, height: 40, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("circle", { r: 5, fill: INK }, b);+ d.push(b);+ }+ return { g, d };+ };+ const ring = (p, w, h) => { const r = el("rect", { fill: "none", stroke: BLUE, "stroke-width": 3, opacity: 0 }, p); r._w = w; r._h = h; return r; };+ const pulse = (r, x, y, k) => {+ const d = 13 * oc(k);+ r.setAttribute("x", (x - r._w / 2 - d).toFixed(1)); r.setAttribute("y", (y - r._h / 2 - d).toFixed(1));+ r.setAttribute("width", (r._w + 2 * d).toFixed(1)); r.setAttribute("height", (r._h + 2 * d).toFixed(1)); r.setAttribute("rx", (r._h / 2 + d).toFixed(1));+ r.setAttribute("opacity", k > 0 && k < 1 ? (0.85 * (1 - k)).toFixed(2) : 0);+ };+ /* where the browser will be: it is not open yet */+ const ph = el("g", {}, svg);+ el("rect", { x: 420, y: 128, width: 336, height: 328, rx: 14, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, ph);+ el("circle", { cx: 588, cy: 292, r: 48, fill: LIL, "fill-opacity": 0.3, stroke: LIL, "stroke-width": 3 }, ph);+ el("path", { d: "M540 292 H636 M588 244 C560 268 560 316 588 340 M588 244 C616 268 616 316 588 340", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, ph);+ el("path", { d: "M420 164 H756", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, ph);+ el("rect", { x: 498, y: 376, width: 180, height: 16, rx: 8, fill: LIL, "fill-opacity": 0.3 }, ph);+ el("rect", { x: 528, y: 404, width: 120, height: 16, rx: 8, fill: LIL, "fill-opacity": 0.3 }, ph);+ /* the app window: step one of the four */+ const app = el("g", {}, svg);+ el("rect", { x: 37, y: 85, width: 370, height: 412, rx: 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, app);+ el("rect", { x: 44, y: 92, width: 356, height: 398, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, app);+ el("path", { d: "M44 124 H400", stroke: INK, "stroke-width": 3 }, app);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, app));+ const step = [0, 1, 2, 3].map((j) => el("circle", { cx: 316 + 20 * j, cy: 108, r: 6, fill: j ? "#fff" : YEL, stroke: INK, "stroke-width": 2.5 }, app));+ const r1 = ring(app, 308, 52), b1 = el("g", {}, app);+ el("rect", { x: -154, y: -26, width: 308, height: 52, rx: 26, fill: YEL, stroke: INK, "stroke-width": 3 }, b1);+ fit(txt(b1, 0, 8, "Open browser to sign in", Object.assign({ "font-size": 22, "font-weight": 700 }, mid)), 272);+ const lab = txt(app, 70, 262, "Code from the browser", { "font-size": 20, fill: MUTE });+ el("rect", { x: 68, y: 276, width: 308, height: 60, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, app);+ const ghost = el("g", {}, app);+ for (let j = 0; j < 6; j++) el("rect", { x: AX - 111 + 38 * j, y: FY - 20, width: 32, height: 40, rx: 7, fill: "none", stroke: LIL, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "1 7" }, ghost);+ const r2 = ring(app, 220, 50), b2 = el("g", {}, app);+ const b2r = el("rect", { x: -110, y: -25, width: 220, height: 50, rx: 25, fill: SOFT, stroke: INK, "stroke-width": 3 }, b2);+ const b2t = txt(b2, 0, 8, "Use this code", Object.assign({ "font-size": 22, "font-weight": 700, fill: MUTE }, mid));+ const done = pill(app, "You are signed in", MINT, true);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 15, YEL);+ button(app, "Watch the sign in again", () => api.seek(0.3));+ /* the browser: more tabs than anyone needs, and the code */+ const B = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -168, y: -164, width: 336, height: 328, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, B);+ el("path", { d: "M-168 -128 V-150 a14 14 0 0 1 14 -14 H154 a14 14 0 0 1 14 14 V-128 Z", fill: SOFT, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, B);+ const tabs = Array.from({ length: 9 }, (_, j) => ({ r: el("rect", { y: -154, height: 26, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2 }, B), c: el("circle", { cy: -141, r: 3.5, fill: [BLUE, MINT, LIL, SKY][j % 4] }, B) }));+ const nt = el("rect", { y: -154, height: 26, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2 }, B);+ el("rect", { x: -150, y: -118, width: 300, height: 22, rx: 11, fill: SOFT }, B);+ const bt = tickDot(B, 17);+ txt(B, 0, -30, "copy the code", Object.assign({ "font-size": 21, fill: MUTE }, mid));+ const sel = el("rect", { x: -121, y: -14, width: 242, height: 56, rx: 10, fill: SKY, opacity: 0 }, B);+ const code = chip(B); code.g.setAttribute("transform", "translate(0 14)");+ el("rect", { x: -100, y: 72, width: 200, height: 14, rx: 7, fill: SOFT }, B);+ el("rect", { x: -70, y: 98, width: 140, height: 14, rx: 7, fill: SOFT }, B);+ const T = chip(svg); mv(T.g); T.g.setAttribute("pointer-events", "none");+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, live = t < 13.2;+ /* press Open browser to sign in */+ tf(b1, AX, 176, (1 - 0.05 * Math.sin(Math.PI * seg(t, 1.3, 1.65)) + (t < 1.3 ? 0.015 * Math.sin(now * 4) : 0)).toFixed(4));+ pulse(r1, AX, 176, seg(t, 1.45, 2.05));+ /* the browser opens where the outline was, and one more tab squeezes in */+ const bo = seg(t, 1.8, 2.45), bs = cl(bk(bo, 1.5), 0, 1.2) * (1 - io(out));+ tf(B, 588, 292, bs.toFixed(4), ((1 - oc(bo)) * -4).toFixed(2));+ B.setAttribute("opacity", bs > 0.02 ? 1 : 0);+ ph.setAttribute("opacity", cl(1 - seg(t, 1.8, 2.1) + seg(t, 13.45, 13.7)).toFixed(2));+ const pitch = lerp(35, 25.2, io(seg(t, 2.35, 3.05)));+ tabs.forEach((o, j) => { o.r.setAttribute("x", (-159 + j * pitch).toFixed(1)); o.r.setAttribute("width", (pitch - 3).toFixed(1)); o.c.setAttribute("cx", (-160.5 + (j + 0.5) * pitch).toFixed(1)); });+ nt.setAttribute("x", (-159 + 9 * pitch).toFixed(1)); nt.setAttribute("width", Math.max(0.1, 317 - 9 * pitch).toFixed(1)); nt.setAttribute("opacity", pitch < 34.6 ? 1 : 0);+ tf(bt, 0, -72, land(seg(t, 2.85, 3.25), 2.4).toFixed(3));+ code.d.forEach((b, j) => tf(b, -95 + 38 * j, 0, land(seg(t, 3.25 + 0.1 * j, 3.55 + 0.1 * j), 2.2).toFixed(3)));+ sel.setAttribute("opacity", (0.9 * seg(t, 4.45, 4.75) * (1 - seg(t, 6.5, 6.9))).toFixed(2));+ /* a copy of the code lifts out, goes across and drops into the field */+ const k = seg(t, 5.1, 6.3), up = oc(seg(t, 4.85, 5.1)) * (1 - io(seg(t, 6.15, 6.4))), tau = t - 6.4;+ tf(T.g, lerp(588, AX, io(k)).toFixed(2), (FY - 11 * up).toFixed(2), (1 + 0.07 * up + (tau > 0 ? 0.05 * Math.exp(-7 * tau) * Math.sin(18 * tau) : 0)).toFixed(4));+ T.g.setAttribute("opacity", t >= 4.85 ? vis.toFixed(2) : 0);+ ghost.setAttribute("opacity", cl(1 - seg(t, 6, 6.3) + out).toFixed(2));+ lab.setAttribute("fill", t >= 5.1 && live ? INK : MUTE);+ /* press Use this code */+ const ready = t >= 6.55 && live;+ b2r.setAttribute("fill", ready ? YEL : SOFT); b2t.setAttribute("fill", ready ? INK : MUTE);+ tf(b2, AX, 389, (1 + 0.06 * Math.sin(Math.PI * seg(t, 6.55, 6.9)) - 0.05 * Math.sin(Math.PI * seg(t, 7.7, 8.05))).toFixed(4));+ pulse(r2, AX, 389, seg(t, 7.85, 8.45));+ const dk = seg(t, 8.45, 8.9);+ tf(done, 210, 452, (land(dk, 2.2) * vis).toFixed(3), ((1 - oc(dk)) * -6).toFixed(2));+ step[0].setAttribute("fill", t >= 8.45 && live ? MINT : YEL);+ tf(sp, 364, 452, (land(seg(t, 8.8, 9.25), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.3 || out > 0 ? "The browser never opens by itself" : t < 2.9 ? "Press Open browser to sign in" : t < 5.1 ? "Copy the code from the browser" : t < 7.4 ? "Paste it into Code from the browser" : t < 8.45 ? "Press Use this code" : "Sign in takes two windows", 700);+ };+ },+ };++ /* ---- fork: a mini golf lane that splits in three, one branch per button; the ball takes the free one and circles the hole first ---- */+ SC.fork = {+ dur: 14, still: 9.2,+ build(svg, api) {+ head(svg, "Run your agents like a company", "tap a branch");+ const LY = [180, 304, 428], HX = 556, FK = 178, A0 = -2.2;+ const BTN = ["Continue with the free version", "I have a licence key", "Get PRO"];+ const DEST = [["Classic view"], ["paste a key"], ["checkout in", "your browser"]];+ const LINE = ["Continue with the free version opens the Classic view", "I have a licence key lets you paste a key", "Get PRO opens the checkout in your browser"];+ const lane = (i) => `M190 304 C250 304 250 ${LY[i]} 310 ${LY[i]} H${HX}`;+ let pick = 0, lastT = 0, held = false;+ /* rough grass, a bunker and a star around the course */+ [[138, 178], [166, 246], [236, 474], [694, 362], [472, 494], [724, 246]].forEach(([x, y]) => el("path", { d: `M${x - 7} ${y} l-3 -11 M${x} ${y} v-14 M${x + 7} ${y} l3 -11`, fill: "none", stroke: MINT, "stroke-width": 3.5, "stroke-linecap": "round" }, svg));+ el("path", { d: "M50 454 C40 422 82 406 112 416 C150 404 176 432 160 460 C148 488 68 488 50 454 Z", fill: YEL, stroke: INK, "stroke-width": 3 }, svg);+ [[84, 440], [112, 456], [136, 438], [96, 468], [140, 464]].forEach(([x, y]) => el("circle", { cx: x, cy: y, r: 2, fill: INK }, svg));+ /* the felt: every lane is drawn twice, ink under mint, so the edge stays one even line */+ const felt = "M64 304 H190 " + [0, 1, 2].map(lane).join(" ");+ el("path", { d: felt, fill: "none", stroke: INK, "stroke-width": 56, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ el("path", { d: felt, fill: "none", stroke: MINT, "stroke-width": 50, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ LY.forEach((y) => el("circle", { cx: HX, cy: y, r: 13, fill: INK }, svg));+ el("rect", { x: 78, y: 287, width: 38, height: 34, rx: 6, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 44, 366, "after setup", { "font-size": 20, fill: MUTE });+ /* one button over each branch; the branch and its button are one tap target */+ const B = BTN.map((s, i) => {+ const g = el("g", {}, svg);+ el("path", { d: lane(i), fill: "none", stroke: "#fff", "stroke-opacity": 0, "stroke-width": 56, "stroke-linecap": "round" }, g);+ const f = el("g", {}, g), t = fit(txt(f, 0, 7, s, { "text-anchor": "middle", "font-size": 19, "font-weight": 700 }), 286);+ const w = Math.round(t.getComputedTextLength() + 38);+ const r = el("rect", { x: -w / 2, y: -20, width: w, height: 40, rx: 20, fill: "#fff", stroke: INK, "stroke-width": 3 });+ f.insertBefore(r, t);+ f.insertBefore(el("rect", { x: -w / 2 - 5, y: -25, width: w + 10, height: 50, rx: 25, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }), r);+ const pr = el("rect", { fill: "none", stroke: BLUE, "stroke-width": 3, opacity: 0 }, g);+ button(g, `Take the branch ${s}`, () => { pick = i; held = true; api.seek(3.4); });+ return { f, r, pr, w, cx: 318 + w / 2, y: LY[i] - 42 };+ });+ const D = DEST.map((L, i) => L.map((s, j) => txt(svg, 604, LY[i] + (L.length > 1 ? -5 + 24 * j : 7), s, { "font-size": 20, "font-weight": 700, fill: MUTE })));+ const disc = [0, 1, 2].map((i) => {+ const g = el("g", { "pointer-events": "none" }, svg);+ el("circle", { r: 16, fill: i ? LIL : MINT, stroke: INK, "stroke-width": 2.5 }, g);+ if (i) el("circle", { r: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ else el("path", { d: "M-7 1 l5 5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ });+ /* the putter, and the ball */+ const put = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M0 0 V190", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, put);+ el("rect", { x: -6, y: -4, width: 12, height: 56, rx: 6, fill: LIL, stroke: INK, "stroke-width": 2.5 }, put);+ el("rect", { x: -10, y: 186, width: 24, height: 27, rx: 5, fill: YEL, stroke: INK, "stroke-width": 2.5 }, put);+ const ball = mv(el("g", { "pointer-events": "none" }, svg));+ el("circle", { r: 11, fill: "#fff", stroke: INK, "stroke-width": 3 }, ball);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 13, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ if (t < lastT - 1 && !held) pick = 0;+ held = false; lastT = t;+ const live = t < 13.2, out = seg(t, 13.2, 13.5), P = pick, Y = LY[P];+ /* the putt */+ const th = t < 0.75 ? 5 * io(seg(t, 0.3, 0.75)) : t < 0.9 ? lerp(5, -2, Math.pow(seg(t, 0.75, 0.9), 2)) : -2 * (1 - oc(seg(t, 0.9, 1.5)));+ tf(put, 62, 104, 1, th.toFixed(2));+ let x = 96, y = 304 + (t < 0.3 ? 0.6 * Math.sin(now * 5) : 0), s = 1, o = 1;+ if (live && t >= 0.88) {+ /* it rolls to the split, stops, and leans toward two branches before it picks */+ const h2 = Math.sin(Math.PI * seg(t, 2.1, 2.7)), h1 = Math.sin(Math.PI * seg(t, 2.8, 3.35));+ x = lerp(96, FK, oc(seg(t, 0.88, 1.9))) + 9 * h2 + 15 * h1; y = 304 + 8 * h2;+ const g = seg(t, 4, 5.9);+ if (g > 0) {+ const BL = P === 1 ? 120 : 181, d = (1 - Math.pow(1 - io(g), 1.6)) * (12 + BL + 237);+ if (d <= 12) x = FK + d;+ else if (d <= 12 + BL) { const q = (d - 12) / BL; x = 190 + 180 * q - 180 * q * q + 120 * q * q * q; y = 304 + (Y - 304) * (3 * q * q - 2 * q * q * q); }+ else { const q = (d - 12 - BL) / 237; x = lerp(310, HX + 16 * Math.cos(A0), q); y = Y + 16 * Math.sin(A0) * io(seg(q, 0.55, 1)); }+ }+ /* round the rim once, then in */+ const k = seg(t, 5.9, 7);+ if (k > 0) { const a = A0 + 2.6 * Math.PI * (1 - Math.pow(1 - k, 1.8)), r = 16 * (1 - Math.pow(k, 2.5)); x = HX + r * Math.cos(a); y = Y + r * Math.sin(a); s = 1 - 0.4 * Math.pow(k, 3); o = 1 - seg(k, 0.82, 1); }+ }+ if (!live) o = seg(t, 13.5, 13.9);+ tf(ball, x.toFixed(2), y.toFixed(2), s.toFixed(3)); ball.setAttribute("opacity", o.toFixed(2));+ let say2 = 0;+ B.forEach((b, i) => {+ const a = 9.9 + 1.6 * (i - 1), pv = live && P === 0 && i > 0 && t >= a && t < a + 1.5;+ if (pv) say2 = i;+ const on = (live && i === P && t >= 3.5) || pv;+ b.r.setAttribute("fill", on ? (i ? LIL : YEL) : "#fff");+ tf(b.f, b.cx, b.y, (1 + 0.05 * Math.sin(Math.PI * seg(t, 0.1 + 0.14 * i, 0.5 + 0.14 * i)) - (i === P ? 0.05 * Math.sin(Math.PI * seg(t, 3.5, 3.85)) : 0) + (P === 0 && i > 0 ? 0.05 * Math.sin(Math.PI * seg(t, a, a + 0.4)) : 0)).toFixed(4));+ const k = i === P ? seg(t, 3.65, 4.25) : 0, e = 13 * oc(k);+ b.pr.setAttribute("x", (b.cx - b.w / 2 - e).toFixed(1)); b.pr.setAttribute("y", (b.y - 20 - e).toFixed(1));+ b.pr.setAttribute("width", (b.w + 2 * e).toFixed(1)); b.pr.setAttribute("height", (40 + 2 * e).toFixed(1)); b.pr.setAttribute("rx", (20 + e).toFixed(1));+ b.pr.setAttribute("opacity", k > 0 && k < 1 ? (0.85 * (1 - k)).toFixed(2) : 0);+ D[i].forEach((n) => n.setAttribute("fill", (live && i === P && t >= 7) || pv ? INK : MUTE));+ tf(disc[i], HX, LY[i], (i === P ? land(seg(t, 6.95, 7.35), 2.2) * (1 - out) : 0).toFixed(3));+ });+ tf(sp, 748, 150, (P === 0 ? land(seg(t, 7.3, 7.75), 2.6) * (1 - out) * (1 + 0.12 * Math.sin(now * 3)) : 0).toFixed(3), (now * 20).toFixed(1));+ say(cap, !live || t < 2 ? "The first screen after setup" : t < 3.5 ? "The screen that stops people" : say2 ? LINE[say2] : P === 0 && t < 7 ? "Without a licence, press Continue with the free version" : LINE[P], 700);+ };+ },+ };++ /* ---- ide: the editor window; an agent adds two lines, Changes marks them, and the switch shows the file as code, split or preview ---- */+ SC.ide = {+ dur: 14, still: 11.6,+ build(svg, api) {+ head(svg, "What is the IDE button for?", "tap the switch");+ const mid = { "text-anchor": "middle" }, PX = 224, PR = 756, PW = PR - PX, TOP = 169.5, BOT = 488.5;+ const LN = (j) => 194 + 25 * j, FY = (j) => 196 + 37 * j, X0 = PX + 48, TW = [360, 236];+ let uv = -1, ut = 0, v0 = 0, last = 0;+ /* the window: title bar, tool row, file list at the side */+ el("rect", { x: 44, y: 92, width: 712, height: 398, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: `M44 124 H756 M44 168 H756 M${PX} 168 V490`, stroke: INK, "stroke-width": 3 }, svg);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 132, y: 97, width: 54, height: 22, rx: 11, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 159, 113.5, "IDE", Object.assign({ "font-size": 16, "font-weight": 700 }, mid));+ el("rect", { x: 262, y: 96, width: 276, height: 24, rx: 12, fill: SOFT }, svg);+ fit(txt(svg, 400, 114, "an agent's working folder", Object.assign({ "font-size": 17 }, mid)), 250);+ let tx = 58;+ const tabs = ["Changes", "History", "Compare"].map((s) => {+ const g = el("g", {}, svg), t = txt(g, 0, 152.5, s, Object.assign({ "font-size": 18, "font-weight": 700 }, mid)), w = Math.round(t.getComputedTextLength() + 26);+ const r = el("rect", { x: tx, y: 131, width: w, height: 30, rx: 9, fill: "#fff", stroke: INK, "stroke-width": 2.5 }), o = { g, r, cx: tx + w / 2 };+ g.insertBefore(r, t); t.setAttribute("x", o.cx); tx += w + 8;+ return o;+ });+ /* the switch: three words, and a yellow knob under the one in use */+ const SW = ["code", "split", "preview"], swg = el("g", {}, svg);+ const swbg = el("rect", { y: 131, height: 30, rx: 15, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, swg), knob = el("rect", { y: 131, height: 30, rx: 15, fill: YEL, stroke: INK, "stroke-width": 2.5 }, swg);+ let sx = 0;+ const S3 = SW.map((s) => { const g = el("g", {}, swg), t = txt(g, 0, 152.5, s, Object.assign({ "font-size": 18, "font-weight": 700 }, mid)), w = Math.round(t.getComputedTextLength() + 28), o = { g, t, w, x: sx }; sx += w; return o; });+ swbg.setAttribute("x", 744 - sx); swbg.setAttribute("width", sx);+ S3.forEach((o, i) => {+ o.x += 744 - sx; o.t.setAttribute("x", o.x + o.w / 2);+ o.g.insertBefore(el("rect", { x: o.x - 4, y: 127, width: o.w + 8, height: 38, rx: 19, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }), o.t);+ button(o.g, `Show the file as ${SW[i]}`, () => { v0 = last; uv = i; ut = api.now(); api.poke(); });+ });+ el("rect", { x: 52, y: FY(2) - 16, width: 164, height: 32, rx: 8, fill: SOFT }, svg);+ [84, 62, 96, 54, 78, 70].forEach((w, j) => {+ const g = el("g", { transform: `translate(${j ? 84 : 72} ${FY(j)})` }, svg);+ if (!j) el("path", { d: "M-9 -7 H-2 L1 -4 H9 V7 H-9 Z", fill: YEL, stroke: INK, "stroke-width": 2.2, "stroke-linejoin": "round" }, g);+ else if (j === 4) { el("rect", { x: -8, y: -7, width: 16, height: 14, rx: 2, fill: SKY, stroke: INK, "stroke-width": 2.2 }, g); el("path", { d: "M-6 5 l5 -6 l4 4 l2 -2 l3 4", fill: "none", stroke: INK, "stroke-width": 1.8, "stroke-linejoin": "round" }, g); }+ else el("path", { d: "M-7 -9 H2 L7 -4 V9 H-7 Z", fill: "#fff", stroke: INK, "stroke-width": 2.2, "stroke-linejoin": "round" }, g);+ el("rect", { x: 18, y: -5, width: w, height: 10, rx: 5, fill: MUTE, "fill-opacity": j === 2 ? 0.85 : 0.4 }, g);+ });+ const dot = el("circle", { r: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("rect", { x: 62, y: 438, width: 144, height: 36, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M82 456 h14 M89 449 v14", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ txt(svg, 106, 462.5, "New file", { "font-size": 18, "font-weight": 700 });+ /* the pane: code on the left of the divider, the formatted preview on the right of it */+ const defs = el("defs", {}, svg);+ const cc = el("rect", { x: PX + 1.5, y: TOP, height: BOT - TOP }, el("clipPath", { id: "mg-ide-code" }, defs)), cp = el("rect", { y: TOP, height: BOT - TOP }, el("clipPath", { id: "mg-ide-prev" }, defs));+ const C = el("g", { "clip-path": "url(#mg-ide-code)" }, svg), V = el("g", { "clip-path": "url(#mg-ide-prev)" }, svg);+ el("rect", { x: PX + 1.5, y: TOP, width: 32, height: BOT - TOP, fill: SOFT }, C);+ const band = [9, 10].map((j) => el("rect", { x: PX + 33.5, y: LN(j) - 12.5, width: PW, height: 25, fill: MINT, "fill-opacity": 0.4, opacity: 0 }, C));+ const mark = [9, 10].map((j) => el("rect", { x: PX + 4, y: LN(j) - 10, width: 6, height: 20, rx: 3, fill: MINT, stroke: INK, "stroke-width": 2, opacity: 0 }, C));+ for (let j = 0; j < 11; j++) el("rect", { x: PX + 15, y: LN(j) - 2, width: 12, height: 4, rx: 2, fill: MUTE, "fill-opacity": 0.45 }, C);+ const bar = (x, j, w, fill, op, h = 10) => el("rect", { x, y: LN(j) - h / 2, width: w, height: h, rx: h / 2, fill, "fill-opacity": op }, C);+ el("rect", { x: X0, y: LN(0) - 6, width: 12, height: 12, rx: 3, fill: LIL }, C); bar(X0 + 20, 0, 150, INK, 0.9, 12);+ bar(X0, 2, 400, MUTE, 0.5); bar(X0, 3, 330, MUTE, 0.5);+ [190, 250, 160].forEach((w, i) => { el("rect", { x: X0, y: LN(5 + i) - 2, width: 9, height: 4, rx: 2, fill: BLUE }, C); bar(X0 + 18, 5 + i, w, MUTE, 0.5); });+ const typed = [bar(X0, 9, 0, INK, 0.75), bar(X0, 10, 0, INK, 0.75)];+ const cur = el("rect", { width: 3, height: 20, fill: INK }, C);+ const pvH = el("rect", { y: 184, height: 20, rx: 10, fill: INK }, V);+ const pvP = [232, 254].map((y) => el("rect", { y, height: 10, rx: 5, fill: MUTE, "fill-opacity": 0.5 }, V));+ const pvL = [298, 324, 350].map((y) => ({ d: el("circle", { cy: y + 5, r: 5, fill: BLUE }, V), b: el("rect", { y, height: 10, rx: 5, fill: MUTE, "fill-opacity": 0.5 }, V) }));+ const pvN = [398, 420].map((y) => el("rect", { y, height: 10, rx: 5, fill: INK, "fill-opacity": 0.75 }, V));+ const div = el("path", { stroke: INK, "stroke-width": 3 }, svg);+ /* the scroll bar: it is a long report */+ el("rect", { x: 740, y: 178, width: 8, height: 302, rx: 4, fill: LIL, "fill-opacity": 0.3 }, svg);+ const thumb = el("rect", { x: 740, y: 181, width: 8, rx: 4, fill: MUTE }, svg);+ const cap = strip(svg, 508, 740);+ const auto = (t) => (t < 7.6 ? io(seg(t, 5.2, 5.85)) : t < 9.8 ? 1 + io(seg(t, 7.6, 8.25)) : t < 13.2 ? 2 - io(seg(t, 9.8, 10.45)) : 1 - io(seg(t, 13.2, 13.7)));+ return (t, now) => {+ const live = t < 13.2, out = seg(t, 13.2, 13.5);+ let v = auto(t);+ if (uv >= 0) { const c = now - ut; if (c >= 5.6) uv = -1; else v = c < 5 ? lerp(v0, uv, io(seg(c, 0, 0.45))) : lerp(uv, v, io(seg(c, 5, 5.6))); }+ last = v;+ const dx = PR - (v / 2) * PW, x = dx + 24, av = Math.max(0, PR - dx - 58);+ cc.setAttribute("width", Math.max(0, dx - PX - 1.5).toFixed(1));+ cp.setAttribute("x", dx.toFixed(1)); cp.setAttribute("width", Math.max(0, PR - 1.5 - dx).toFixed(1));+ div.setAttribute("d", `M${dx.toFixed(1)} 168 V${(dx > PR - 14 ? 476 + Math.sqrt(196 - Math.pow(dx - PR + 14, 2)) : 490).toFixed(1)}`);+ const a = Math.min(1, Math.floor(v)), f = v - a;+ knob.setAttribute("x", lerp(S3[a].x, S3[a + 1].x, f).toFixed(1)); knob.setAttribute("width", lerp(S3[a].w, S3[a + 1].w, f).toFixed(1));+ /* the agent types two more lines, and the file in the list gets its dot */+ const ty = [seg(t, 0.9, 1.7), seg(t, 1.8, 2.4)].map((k) => (Math.round(k * 12) / 12) * (1 - out)), ci = t >= 1.8 ? 1 : 0;+ typed.forEach((n, i) => n.setAttribute("width", (TW[i] * ty[i]).toFixed(1)));+ cur.setAttribute("x", (X0 + TW[ci] * ty[ci] + 4).toFixed(1)); cur.setAttribute("y", LN(9 + ci) - 10);+ cur.setAttribute("opacity", Math.sin(now * 6) > -0.2 ? 1 : 0);+ tf(dot, 204, FY(2), (land(seg(t, 0.95, 1.35), 2.4) * (1 - out)).toFixed(3));+ thumb.setAttribute("height", (15 - 2 * (ty[0] + ty[1])).toFixed(1));+ /* Changes lights up and marks the two new lines */+ tabs[0].r.setAttribute("fill", live && t >= 3.2 ? YEL : "#fff");+ tabs[0].g.setAttribute("transform", `translate(${tabs[0].cx} 146) scale(${(1 + 0.09 * Math.sin(Math.PI * seg(t, 3.2, 3.6))).toFixed(4)}) translate(${-tabs[0].cx} -146)`);+ band.forEach((n, i) => { const k = (seg(t, 3.5 + 0.2 * i, 3.8 + 0.2 * i) * (1 - out)).toFixed(2); n.setAttribute("opacity", k); mark[i].setAttribute("opacity", k); });+ /* the preview keeps its own layout, whatever room it has */+ pvH.setAttribute("x", x.toFixed(1)); pvH.setAttribute("width", (0.62 * av).toFixed(1));+ pvP.forEach((n, i) => { n.setAttribute("x", x.toFixed(1)); n.setAttribute("width", ([1, 0.78][i] * av).toFixed(1)); });+ pvL.forEach((o, i) => { o.d.setAttribute("cx", (x + 6).toFixed(1)); o.b.setAttribute("x", (x + 22).toFixed(1)); o.b.setAttribute("width", ([0.55, 0.72, 0.45][i] * Math.max(0, av - 22)).toFixed(1)); });+ pvN.forEach((n, i) => { n.setAttribute("x", x.toFixed(1)); n.setAttribute("width", ([0.9, 0.58][i] * av * ty[i]).toFixed(1)); });+ say(cap, uv >= 0 || (t >= 5.2 && t < 9.8) ? "Preview Markdown as code, split or preview" : !live || t < 0.9 ? "The IDE button opens a code editor on an agent's working folder" : t < 3.2 ? "See what an agent just did without leaving the office" : t < 5.2 ? "See what the agent changed: Changes, History and Compare" : "Read the report in preview next to the files it describes", 700);+ };+ },+ };++ /* ---- fixes: four stepping stones over a stream, one per fix, each a bigger restart than the last; the pebble stops on the first that works ---- */+ SC.fixes = {+ dur: 14, still: 11.6,+ build(svg, api) {+ head(svg, "What do you do when an install fails?", "tap a stone");+ const mid = { "text-anchor": "middle" }, T0 = 0.9, GAP = 2, B0 = [68, 356];+ const ST = [[176, 344, 56, 36], [312, 326, 60, 39], [456, 308, 64, 42], [608, 290, 68, 45]];+ const NAME = [["Check again"], ["restart", "the agent"], ["reopen", "the app"], ["restart the", "computer"]];+ const STEP = ["Press Check again or Try again", "Close that agent and start it again", "Quit Munder Difflin completely and open it again", "Restart the computer"];+ const LP = ST.map(([x, y, rx]) => [x - rx + 22, y + 2]);+ const tick = (p) => el("path", { d: "M-6 1 l4.5 4.5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, p);+ let goal = 2, lastT = 0, held = false;+ /* the stream: ripples, lily pads, and a bank on each side */+ const rip = [[150, 250], [268, 244], [392, 232], [512, 224], [150, 446], [304, 458], [566, 442], [644, 474], [440, 482]].map(([x, y]) => ({ x, y, n: el("path", { d: "M-16 0 q8 -7 16 0 t16 0", fill: "none", stroke: SKY, "stroke-width": 3.5, "stroke-linecap": "round" }, svg) }));+ const pad = [[230, 470, 15], [376, 448, 13], [506, 458, 17], [708, 462, 14]].map(([x, y, r]) => ({ x, y, n: el("path", { d: `M0 0 L${r} 0 A${r} ${r} 0 1 1 ${(r * 0.8).toFixed(1)} ${(-r * 0.6).toFixed(1)} Z`, fill: MINT, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg) }));+ el("path", { d: "M36 336 C32 304 76 290 94 310 C112 330 108 392 88 408 C68 424 42 412 36 386 Z", fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M696 300 C702 266 750 262 764 298 C772 330 768 390 748 406 C726 420 700 398 694 364 Z", fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ [[54, 330], [84, 394], [722, 302], [746, 378], [718, 352]].forEach(([x, y]) => el("path", { d: `M${x - 5} ${y} l-2 -8 M${x} ${y} v-10 M${x + 5} ${y} l2 -8`, fill: "none", stroke: INK, "stroke-width": 2.2, "stroke-linecap": "round" }, svg));+ /* the stones: the same restart arrow on each, a size bigger every time, around the thing it restarts */+ const S = ST.map(([x, y, rx, ry], k) => {+ const g = el("g", {}, svg), R = [18, 20, 22, 25][k];+ el("ellipse", { cx: x, cy: y, rx: rx + 10, ry: ry + 8, fill: "none", stroke: SKY, "stroke-width": 3 }, g);+ const rp = el("ellipse", { cx: x, cy: y, fill: "none", stroke: SKY, "stroke-width": 3, opacity: 0 }, g);+ el("ellipse", { cx: x, cy: y, rx: rx + 5, ry: ry + 5, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const e = el("ellipse", { cx: x, cy: y, rx, ry, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ const io2 = el("g", { transform: `translate(${x + 15} ${y})` }, g), sp = el("g", {}, io2), ic = el("g", { transform: `scale(${1.15 + 0.05 * k})` }, io2), ex = (R * 0.5).toFixed(1), ey = (-R * 0.866).toFixed(1);+ el("path", { d: `M${R} 0 A${R} ${R} 0 1 1 ${ex} ${ey} M${ex} ${ey} l-3 -6.3 M${ex} ${ey} l-7 0.6`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, sp);+ if (k === 0) el("rect", { x: -9, y: -5, width: 18, height: 10, rx: 5, fill: YEL, stroke: INK, "stroke-width": 2.2 }, ic);+ if (k === 1) { el("rect", { x: -10, y: -8, width: 20, height: 16, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M-5 -3 l4 3 l-4 3 M1 4 h5", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-linejoin": "round" }, ic); }+ if (k === 2) { el("rect", { x: -12, y: -9, width: 24, height: 18, rx: 3, fill: LIL, stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M-12 -3 H12", stroke: INK, "stroke-width": 2.2 }, ic); }+ if (k === 3) { el("rect", { x: -14, y: -12, width: 28, height: 19, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M0 7 V12 M-8 12.5 H8", stroke: INK, "stroke-width": 2.2, "stroke-linecap": "round" }, ic); }+ NAME[k].forEach((s, j) => txt(svg, x, y + ry + 28 + 23 * j, s, Object.assign({ "font-size": 20, "font-weight": 700 }, mid)));+ button(g, `Hop as far as ${NAME[k].join(" ")}`, () => { goal = k; held = true; api.seek(0.7); });+ return { e, sp, rp, rx, ry, bx: x + 0.707 * rx, by: y - 0.707 * ry };+ });+ S.forEach((s) => {+ s.bd = el("g", { "pointer-events": "none" }, svg);+ s.no = el("g", {}, s.bd); el("circle", { r: 11, fill: LIL, stroke: INK, "stroke-width": 2.5 }, s.no); el("path", { d: "M-4 -4 L4 4 M4 -4 L-4 4", stroke: INK, "stroke-width": 2.8, "stroke-linecap": "round" }, s.no);+ s.ok = el("g", {}, s.bd); el("circle", { r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, s.ok); tick(s.ok).setAttribute("transform", "scale(0.85)");+ });+ /* the install that failed: a crooked card with a broken bar */+ const card = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: -108, y: -56, width: 216, height: 112, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, card);+ txt(card, -88, -14, "Install", { "font-size": 24, "font-weight": 700 });+ el("rect", { x: -88, y: 10, width: 96, height: 18, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, card);+ const brk = el("g", {}, card); el("rect", { x: 0, y: -9, width: 76, height: 18, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, brk);+ const mend = el("rect", { x: -88, y: 10, width: 176, height: 18, rx: 6, fill: MINT, stroke: INK, "stroke-width": 2.5, opacity: 0 }, card);+ const warn = el("g", {}, card); el("circle", { r: 14, fill: YEL, stroke: INK, "stroke-width": 2.5 }, warn); el("path", { d: "M0 -7 V2", stroke: INK, "stroke-width": 3.2, "stroke-linecap": "round" }, warn); el("circle", { cy: 7.5, r: 2, fill: INK }, warn);+ const okm = el("g", {}, card); el("circle", { r: 14, fill: MINT, stroke: INK, "stroke-width": 2.5 }, okm); tick(okm);+ /* past the far bank: the last resort */+ const bub = el("g", { transform: "translate(650 154)", "pointer-events": "none" }, svg);+ el("rect", { x: -114, y: -50, width: 228, height: 100, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 3 }, bub);+ el("path", { d: "M40 48 L58 80 L76 48", fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round", "stroke-linecap": "round" }, bub);+ txt(bub, 0, -12, "Still stuck?", Object.assign({ "font-size": 20, fill: MUTE }, mid));+ fit(txt(bub, 0, 20, "Ask your orchestrator", Object.assign({ "font-size": 20, "font-weight": 700 }, mid)), 204);+ const sh = el("ellipse", { rx: 10, ry: 4, fill: INK, opacity: 0.16, "pointer-events": "none" }, svg);+ const peb = mv(el("g", { "pointer-events": "none" }, svg));+ el("circle", { r: 11, fill: YEL, stroke: INK, "stroke-width": 3 }, peb); el("circle", { cx: -3.5, cy: -3.5, r: 2.5, fill: "#fff" }, peb);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 12, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ if (t < lastT - 1 && !held) goal = 2;+ held = false; lastT = t;+ const live = t < 13.2, out = seg(t, 13.2, 13.5), Vg = T0 + GAP * goal + 1.35;+ rip.forEach((o, i) => tf(o.n, (o.x + 4 * Math.sin(now * 0.8 + i * 1.7)).toFixed(2), o.y));+ pad.forEach((o, i) => tf(o.n, (o.x + 5 * Math.sin(now * 0.4 + i * 2)).toFixed(2), (o.y + 2 * Math.cos(now * 0.5 + i)).toFixed(2), 1, (20 * i + 8 * Math.sin(now * 0.3 + i)).toFixed(1)));+ /* the pebble: one hop per stone, as far as the stone that works */+ let px = B0[0], py = B0[1], lift = 0, tl = -1, po = 1;+ for (let k = 0; k <= goal; k++) {+ const h = T0 + GAP * k; if (t < h) break;+ const a = k ? LP[k - 1] : B0, q = seg(t, h, h + 0.5), e = 0.5 * q + 0.5 * io(q);+ px = lerp(a[0], LP[k][0], e); py = lerp(a[1], LP[k][1], e); lift = 4 * q * (1 - q); tl = t - h - 0.5;+ }+ if (!live) { po = t < 13.5 ? 1 - out : seg(t, 13.55, 13.9); if (t >= 13.5) { px = B0[0]; py = B0[1]; lift = 0; tl = -1; } }+ const sq = tl > 0 ? 0.2 * Math.exp(-9 * tl) * Math.cos(20 * tl) : 0, bob = tl > 1 ? 1.2 * Math.sin(now * 2.2) : 0, ps = 1 + 0.22 * lift;+ peb.setAttribute("transform", `translate(${px.toFixed(2)} ${(py - 42 * lift + bob + 11 * sq).toFixed(2)}) scale(${(ps * (1 + 0.5 * sq)).toFixed(4)} ${(ps * (1 - sq)).toFixed(4)})`);+ peb.setAttribute("opacity", po.toFixed(2));+ tf(sh, px.toFixed(2), (py + 10).toFixed(2), (1 - 0.45 * lift).toFixed(3)); sh.setAttribute("opacity", (0.16 * po).toFixed(3));+ S.forEach((s, k) => {+ const h = T0 + GAP * k, used = k <= goal, v = h + 1.35, r = used ? seg(t, h + 0.5, h + 1.25) : 0;+ s.e.setAttribute("fill", used && live && t >= v ? (k === goal ? MINT : SOFT) : "#fff");+ s.sp.setAttribute("transform", `rotate(${used ? (360 * io(seg(t, h + 0.6, h + 1.3))).toFixed(1) : 0})`);+ s.rp.setAttribute("rx", (s.rx + 10 + 20 * oc(r)).toFixed(1)); s.rp.setAttribute("ry", (s.ry + 8 + 14 * oc(r)).toFixed(1)); s.rp.setAttribute("opacity", r > 0 && r < 1 ? (0.9 * (1 - r)).toFixed(2) : 0);+ tf(s.bd, s.bx.toFixed(1), s.by.toFixed(1), (used ? land(seg(t, v, v + 0.4), 2.4) * (1 - out) : 0).toFixed(3));+ s.no.setAttribute("opacity", k === goal ? 0 : 1); s.ok.setAttribute("opacity", k === goal ? 1 : 0);+ });+ /* each try lifts the broken half of the bar; only the last one holds */+ let th = -5 + 0.5 * Math.sin(now * 1.6), ph = 10;+ for (let k = 0; k < goal; k++) { const a = seg(t, T0 + GAP * k + 0.6, T0 + GAP * k + 1.55); if (a > 0 && a < 1) { ph = 10 - 7 * Math.sin(Math.PI * a); th = -5 + 1.8 * Math.sin(a * Math.PI * 4) * (1 - a); } }+ if (t >= Vg - 0.75) { ph = 10 * (1 - io(seg(t, Vg - 0.75, Vg - 0.05))); if (t >= Vg) th = -5 * (1 - land(seg(t, Vg, Vg + 0.6), 1.8)); }+ const bk2 = io(out), fixed = live && t >= Vg;+ tf(card, 160, 158, 1, lerp(th, -5, bk2).toFixed(2)); tf(brk, 12, 19, 1, lerp(ph, 10, bk2).toFixed(2));+ mend.setAttribute("opacity", (seg(t, Vg, Vg + 0.25) * (1 - out)).toFixed(2));+ tf(warn, 78, -24, (fixed ? 0 : (1 + 0.07 * Math.sin(now * 5)) * (live ? 1 : seg(t, 13.3, 13.6))).toFixed(3));+ tf(okm, 78, -24, (land(seg(t, Vg, Vg + 0.4), 2.4) * (1 - out)).toFixed(3));+ tf(sp, 284, 100, (land(seg(t, Vg + 0.3, Vg + 0.75), 2.6) * (1 - out) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, !live || t < T0 ? "When an install fails, work up four steps" : t >= Vg ? "Stop at the first step that fixes it" : STEP[Math.min(goal, Math.floor((t - T0) / GAP))], 700);+ };+ },+ };++ /* ==== ADD NEW SCENES ABOVE THIS LINE ==== */++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/office.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/office.pngnew file mode 100644index 00000000..5d1a2239Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/office.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.pngnew file mode 100644index 00000000..edd4c53eBinary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/rooms.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/rooms.pngnew file mode 100644index 00000000..95a65f88Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/rooms.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/signin.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/signin.pngnew file mode 100644index 00000000..75cb6263Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/signin.png differdiff --git a/blog/src/assets/media/how-to-install-and-use-munder-difflin/terminal.png b/blog/src/assets/media/how-to-install-and-use-munder-difflin/terminal.pngnew file mode 100644index 00000000..0a5c6e42Binary files /dev/null and b/blog/src/assets/media/how-to-install-and-use-munder-difflin/terminal.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/ask-me-056.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/ask-me-056.pngnew file mode 100644index 00000000..2da60968Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/ask-me-056.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/askme.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/askme.pngnew file mode 100644index 00000000..5c9e159cBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/askme.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/brief.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/brief.pngnew file mode 100644index 00000000..b377d970Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/brief.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/hero.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/hero.pngindex 6eb3ee69..f2166cc6 100644Binary files a/blog/src/assets/media/your-first-hour-with-munder-difflin/hero.png and b/blog/src/assets/media/your-first-hour-with-munder-difflin/hero.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/hour.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/hour.pngnew file mode 100644index 00000000..40dd9497Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/hour.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.css b/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.cssnew file mode 100644index 00000000..859374ca--- /dev/null+++ b/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.js b/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.jsnew file mode 100644index 00000000..aa644935--- /dev/null+++ b/blog/src/assets/media/your-first-hour-with-munder-difflin/motion.js@@ -0,0 +1,1247 @@+/* Bright coded scenes for "Your First Hour With Munder Difflin 0.5.6". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.++ TO ADD A SCENE: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } just above+ the line marked "ADD NEW SCENES ABOVE THIS LINE". Change nothing else in this file.+ - The drawing is 800 by 600 units (shown 1600 by 1200). dur is the loop in seconds, still is the frame for NAME.png.+ - build draws everything once and returns the frame function: t is the loop time, now never wraps (idle motion).+ - api.seek(s) jumps the loop to second s (a replay). api.now() is the clock. api.poke() redraws after a tap.+ - Helpers: el, txt, head (heading plus tap hint), strip and say (the caption line), button (tap and keyboard),+ pill, star, fit, mv (marks a group that moves, for the measurement tests), seg, io, oc, bk, land, lerp, tf. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };+ /* a status pill sized to its word, with or without a tick */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* mark a group as moving, so the measurement tests can see when it crosses a label */+ const mv = (g) => { g.setAttribute("data-mv", ""); return g; };++ const SC = {};++ /* ---- title card, 16 by 9: a stopwatch whose hand goes once round, one hour, and then it rings ---- */+ SC.hero = {+ w: 960, h: 540, dur: 4, still: 2.6,+ build(svg) {+ txt(svg, 66, 92, "From first sign in to a working office", { "font-size": 25, "font-weight": 500, fill: MUTE });+ const L = ["Your First Hour", "With Munder Difflin", "0.5.6"].map((s, i) => txt(svg, 62, 176 + i * 74, s, { "font-size": 76, "font-weight": 700 }));+ const w = Math.max(...L.map((n) => n.getComputedTextLength()));+ if (w > 556) L.forEach((n) => n.setAttribute("font-size", Math.floor((76 * 556) / w)));+ el("path", { d: `M68 344 H${Math.round(56 + L[2].getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ const pt = txt(svg, 0, 419, "Onboarding guide", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 }), pw = Math.round(pt.getComputedTextLength() + 64);+ svg.insertBefore(el("rect", { x: 62, y: 382, width: pw, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }), pt);+ pt.setAttribute("x", 62 + pw / 2);+ el("path", { d: "M70 494 C190 468 300 518 420 490 S560 492 620 478", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 494, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ /* the stopwatch: a ring on top, two side buttons, the crown, a face with sixty marks */+ const c = el("g", {}, svg);+ el("circle", { cx: 0, cy: -172, r: 15, fill: "none", stroke: INK, "stroke-width": 4 }, c);+ [-1, 1].forEach((s) => el("rect", { x: -12, y: -142, width: 24, height: 26, rx: 6, fill: LIL, stroke: INK, "stroke-width": 3, transform: `rotate(${s * 40})` }, c));+ el("rect", { x: -11, y: -136, width: 22, height: 22, fill: "#fff", stroke: INK, "stroke-width": 3 }, c);+ const crown = el("rect", { x: -24, y: -160, width: 48, height: 26, rx: 7, fill: YEL, stroke: INK, "stroke-width": 3 }, c);+ el("circle", { r: 118, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ el("circle", { r: 98, fill: "#fff", stroke: INK, "stroke-width": 3 }, c);+ const LEN = 2 * Math.PI * 62;+ const ring = el("circle", { r: 62, fill: "none", stroke: YEL, "stroke-width": 22, transform: "rotate(-90)" }, c);+ el("circle", { r: 88, fill: "none", stroke: MUTE, "stroke-width": 7, "stroke-dasharray": `2 ${((2 * Math.PI * 88) / 60 - 2).toFixed(3)}`, "stroke-dashoffset": 1 }, c);+ for (let i = 0; i < 12; i++) el("path", { d: "M0 -76 V-92", stroke: INK, "stroke-width": 4.5, "stroke-linecap": "round", transform: `rotate(${i * 30})` }, c);+ const hand = el("path", { d: "M0 16 V-78", stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, c);+ el("circle", { r: 10, fill: INK }, c);+ el("circle", { r: 4, fill: YEL }, c);+ const ding = el("path", { d: [-1, 1].map((s) => `M${s * 128} -42 q${s * 12} 42 0 84 M${s * 142} -60 q${s * 14} 60 0 120`).join(" "), fill: "none", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, c);+ star(svg, 646, 132, 15, YEL); star(svg, 912, 470, 17, MINT); star(svg, 662, 456, 12, SKY);+ return (t) => {+ const k = io(seg(t, 0.35, 2.3)), out = io(seg(t, 3.5, 3.95)), tau = t - 2.3, ring2 = tau > 0 && tau < 1.3 ? Math.exp(-2.5 * tau) : 0;+ ring.setAttribute("stroke-dasharray", `${(LEN * k * (1 - out)).toFixed(2)} ${(LEN + 4).toFixed(2)}`);+ hand.setAttribute("transform", `rotate(${(360 * k * (1 - out) + (tau > 0 ? 6 * Math.exp(-5 * tau) * Math.sin(20 * tau) : 0)).toFixed(2)})`);+ crown.setAttribute("transform", `translate(0 ${(5 * Math.sin(Math.PI * seg(t, 0.1, 0.35))).toFixed(2)})`);+ c.setAttribute("transform", `translate(${(794 + 2.2 * ring2 * Math.sin(tau * 60)).toFixed(2)} 296)`);+ ding.setAttribute("opacity", (seg(t, 2.3, 2.4) * (1 - seg(t, 3.3, 3.55))).toFixed(2));+ ding.setAttribute("transform", `scale(${(1 + 0.04 * Math.sin(Math.max(0, tau) * 26)).toFixed(3)})`);+ };+ },+ };++ /* ---- hour: a bus route with six stops, one for each part of the hour; the bus takes the hairpin a little fast ---- */+ SC.hour = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Your first hour", "tap a stop");+ const X0 = 122, XE = 634, R1 = 222, R2 = 406, RAD = 92, RUN = XE - X0, ARC = Math.PI * RAD, LEN = 2 * RUN + ARC;+ const S = [0, 216, 432, RUN + ARC + 80, RUN + ARC + 296, LEN];+ const LIT = [0.8, 2.6, 4.6, 7.3, 9.3, 11.3], DEP = [1.6, 3.5, 5.5, 8.2, 10.2];+ const ST = [["Minute 0", ["sign in"]], ["Minute 10", ["a first job"]], ["Minute 20", ["a small team"]], ["Minute 35", ["work that starts", "without you"]], ["Minute 45", ["temps and tokens"]], ["Minute 50", ["the Pro screens"]]];+ const SAY = ["Minute 0: sign in", "Minute 10: give your orchestrator one real job", "Minute 20: hire a small team", "Minute 35: set up work that starts without you", "Minute 45: temps and tokens", "Minute 50: learn the Pro screens"];+ /* where the bus is at distance s along the route, and how far round the hairpin it has come */+ const at = (s) => {+ if (s <= RUN) return [X0 + s, R1, 0];+ if (s <= RUN + ARC) { const a = (s - RUN) / RAD; return [XE + RAD * Math.sin(a), R1 + RAD - RAD * Math.cos(a), a]; }+ return [XE - (s - RUN - ARC), R2, Math.PI];+ };+ /* trees and a pond, because it is a map */+ const trees = [[700, 164, 17], [744, 160, 23], [668, 494, 15], [710, 498, 23], [752, 492, 17]].map(([x, y, r]) => {+ const g = el("g", {}, svg);+ el("path", { d: `M${x} ${y} V${y - r - 10}`, stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round" }, g);+ el("circle", { cx: x, cy: y - r - 14, r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ return { g, x, y };+ });+ el("ellipse", { cx: 662, cy: 314, rx: 36, ry: 24, fill: SKY, stroke: INK, "stroke-width": 2.5 }, svg);+ const rip = el("ellipse", { cx: 662, cy: 314, fill: "none", stroke: "#fff", "stroke-width": 3 }, svg);+ /* the road, and the part of it the bus has already driven */+ const D = `M${X0} ${R1} H${XE} A${RAD} ${RAD} 0 0 1 ${XE} ${R2} H${X0}`;+ el("path", { d: D, fill: "none", stroke: INK, "stroke-width": 18, "stroke-linecap": "round" }, svg);+ el("path", { d: D, fill: "none", stroke: "#fff", "stroke-width": 12, "stroke-linecap": "round" }, svg);+ const done = el("path", { d: D, fill: "none", stroke: BLUE, "stroke-width": 12, "stroke-linecap": "round" }, svg);+ /* skid marks on the outside of the hairpin */+ const A0 = 0.95, A1 = 2.2;+ const skid = [RAD + 16, RAD + 23].map((r) => {+ const p = (a) => `${(XE + r * Math.sin(a)).toFixed(1)} ${(R1 + RAD - r * Math.cos(a)).toFixed(1)}`;+ return { n: el("path", { d: `M${p(A0)} A${r} ${r} 0 0 1 ${p(A1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg), L: r * (A1 - A0) };+ });+ /* the six stops; each one is a button that sends the bus there */+ let ov = null, cur = 0;+ const send = (i) => { ov = { s: cur, to: S[i], t0: api.now() }; api.seek(LIT[i] - 0.8); };+ const stops = ST.map(([m, nm], i) => {+ const [x, y] = at(S[i]), g = el("g", {}, svg);+ el("rect", { x: x - 102, y: y - 112, width: 204, height: nm.length > 1 ? 192 : 162, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const p = pill(g, m, SOFT);+ const dot = el("circle", { cx: x, cy: y, r: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ const names = nm.map((s, j) => txt(g, x, y + 44 + j * 27, s, { "text-anchor": "middle", "font-size": 22, fill: MUTE }));+ button(g, `Send the bus to ${m.toLowerCase()}: ${nm.join(" ")}`, () => send(i));+ return { x, y, p, r: p.firstChild, dot, names };+ });+ /* the bus, side on; it stays upright and turns round in the hairpin */+ const bus = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -35, y: -46, width: 70, height: 34, rx: 9, fill: YEL, stroke: INK, "stroke-width": 3 }, bus);+ [-28, -10].forEach((x) => el("rect", { x, y: -40, width: 14, height: 12, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2 }, bus));+ el("rect", { x: 8, y: -40, width: 21, height: 12, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2 }, bus);+ el("path", { d: "M-35 -21 H35", stroke: INK, "stroke-width": 2 }, bus);+ const wh = [-18, 18].map((x) => {+ const w = el("g", {}, bus);+ el("circle", { r: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, w);+ el("path", { d: "M-4 0 H4", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, w);+ return { w, x };+ });+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 13, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.45), vis = 1 - out, home = t >= 13.45;+ let s = home ? 0 : LEN, last = 0, turn = 1;+ for (let i = 0; i < 6; i++) if (t >= LIT[i]) last = i;+ if (!home) for (let i = 0; i < 5; i++) if (t < LIT[i + 1]) { s = lerp(S[i], S[i + 1], io(seg(t, DEP[i], LIT[i + 1]))); break; }+ /* after a tap the bus drives from where it was to the stop that was tapped, then turns to face the route again */+ if (ov) {+ const k = (now - ov.t0) / 0.8;+ if (k < 0 || k >= 1) ov = null;+ else { const d = Math.sign(ov.to - ov.s) || 1; s = lerp(ov.s, s, io(k)); turn = lerp(d, 1, io(seg(k, 0.72, 1))); }+ }+ cur = s;+ const [x, y, a] = at(s);+ const face = lerp(1, -1, io(seg(a, 1.15, 2.0))) * turn, lean = 13 * Math.sin(a);+ const ta = t - LIT[last], hop = !ov && ta > 0 && ta < 1.2 ? 4 * Math.exp(-6 * ta) * Math.abs(Math.sin(15 * ta)) : 0;+ const idle = t > 11.9 && !home ? 0.8 * Math.sin(now * 14) : 0;+ bus.setAttribute("transform", `translate(${x.toFixed(2)} ${(y - 8 - hop + idle).toFixed(2)}) rotate(${lean.toFixed(2)}) scale(${face.toFixed(3)} 1)`);+ bus.setAttribute("opacity", (home ? seg(t, 13.5, 13.75) : vis).toFixed(2));+ wh.forEach((w) => w.w.setAttribute("transform", `translate(${w.x} -8) rotate(${((s / 8) * 57.3).toFixed(1)})`));+ done.setAttribute("stroke-dasharray", `${s.toFixed(1)} ${LEN + 40}`);+ done.setAttribute("opacity", vis.toFixed(2));+ const sk = seg(s, RUN + RAD * (A0 + 0.2), RUN + RAD * (A1 + 0.35));+ skid.forEach((m) => { m.n.setAttribute("stroke-dasharray", `${(m.L * sk).toFixed(1)} ${(m.L + 20).toFixed(1)}`); m.n.setAttribute("opacity", sk > 0 ? vis.toFixed(2) : 0); });+ stops.forEach((st, i) => {+ const lit = t >= LIT[i] && !home && out < 0.5, bump = Math.sin(Math.PI * seg(t, LIT[i], LIT[i] + 0.4)) * vis;+ tf(st.p, st.x, st.y - 84, (1 + 0.12 * bump).toFixed(3));+ st.r.setAttribute("fill", lit ? YEL : SOFT);+ st.dot.setAttribute("fill", lit ? YEL : "#fff");+ st.dot.setAttribute("r", (12 + 3 * bump).toFixed(2));+ st.names.forEach((n) => n.setAttribute("fill", lit ? INK : MUTE));+ });+ tf(sp, 182, 366, (land(seg(t, 11.5, 11.9), 2.6) * (home ? 0 : vis) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ trees.forEach((tr, i) => tr.g.setAttribute("transform", `rotate(${(1.6 * Math.sin(now * 1.4 + i * 1.3)).toFixed(2)} ${tr.x} ${tr.y})`));+ const rp = (now * 0.4 + 0.35) % 1;+ rip.setAttribute("rx", (6 + 24 * rp).toFixed(1)); rip.setAttribute("ry", (4 + 15 * rp).toFixed(1)); rip.setAttribute("opacity", (1 - rp).toFixed(2));+ say(cap, t >= LIT[0] && !home && out < 0.5 ? SAY[last] : "The first hour as a route", 700);+ };+ },+ };++ /* ---- brief: a long toaster with two slots; a vague note comes back as a question, a full brief as three tasks ---- */+ SC.brief = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Give your orchestrator a first job", "tap a note");+ const XA = 212, XB = 588, TOP = 388;+ const clip = el("clipPath", { id: "mg-brief-slots" }, el("defs", {}, svg));+ el("rect", { x: 0, y: 0, width: 800, height: TOP + 1 }, clip);+ /* the two notepads; each is a button that drops its note again */+ const pad = (x, tag, tagFill, lines, label, at) => {+ const g = el("g", {}, svg);+ el("rect", { x: x - 152, y: 80, width: 304, height: 190, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ el("rect", { x: x - 126, y: 134, width: 264, height: 126, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: x - 132, y: 128, width: 264, height: 126, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: x - 132, y: 128, width: 264, height: 20, rx: 8, fill: LIL, stroke: INK, "stroke-width": 3 }, g);+ [0, 1, 2].forEach((j) => el("path", { d: `M${x - 114} ${189 + j * 28} H${x + 114}`, stroke: LIL, "stroke-width": 1.5 }, g));+ lines.forEach((s, j) => {+ el("circle", { cx: x - 108, cy: 175 + j * 28, r: 4.5, fill: BLUE }, g);+ txt(g, x - 96, 182 + j * 28, s, { "font-size": 21, "font-weight": 600 });+ });+ tf(pill(g, tag, tagFill), x, 104);+ button(g, label, () => api.seek(at));+ };+ pad(XA, "Weak", SOFT, ["clean up my project"], "Drop the weak note again", 0.7);+ pad(XB, "Better", MINT, ["the outcome", "the folder", "what done looks like"], "Drop the better note again", 5.0);+ /* the back of the toaster: the two slots, which glow while it works */+ const back = el("g", {}, svg);+ const slots = [XA, XB].map((x) => el("rect", { x: x - 80, y: TOP - 9, width: 160, height: 18, rx: 9, fill: INK, stroke: INK, "stroke-width": 3 }, back));+ /* what goes in and what comes out, clipped at the top of the toaster */+ const lay = el("g", { "clip-path": "url(#mg-brief-slots)" }, svg);+ const slip = (n) => {+ const g = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -36, y: -26, width: 72, height: 52, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: -36, y: -26, width: 72, height: 11, rx: 5, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: n === 1 ? "M-22 4 H16" : "M-22 -2 H20 M-22 8 H12 M-22 18 H22", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ };+ const SA = slip(1), SB = slip(3);+ const Q = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -52, y: -116, width: 104, height: 140, rx: 10, fill: SOFT, stroke: INK, "stroke-width": 3 }, Q);+ txt(Q, 0, -38, "?", { "text-anchor": "middle", "font-size": 70, "font-weight": 700, fill: MUTE });+ const tasks = [SKY, LIL, YEL].map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -38, y: -98, width: 76, height: 94, rx: 8, fill: c, stroke: INK, "stroke-width": 3 }, g);+ const box = el("rect", { x: -26, y: -86, width: 24, height: 24, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ const tick = el("path", { d: "M-20 -74 l4.5 5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ el("path", { d: "M6 -74 H26 M-26 -48 H26 M-26 -33 H26 M-26 -18 H10", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, box, tick };+ });+ /* the front of the toaster */+ const front = el("g", {}, svg);+ const BOT = TOP + 102;+ el("path", { d: `M132 ${TOP} H668 A40 40 0 0 1 708 ${TOP + 40} V${BOT - 14} A14 14 0 0 1 694 ${BOT} H106 A14 14 0 0 1 92 ${BOT - 14} V${TOP + 40} A40 40 0 0 1 132 ${TOP} Z`, fill: LIL, stroke: INK, "stroke-width": 3 }, front);+ el("path", { d: `M136 ${TOP + 20} H664`, stroke: "#fff", "stroke-width": 4, "stroke-linecap": "round" }, front);+ [122, 644].forEach((x) => el("rect", { x, y: BOT, width: 34, height: 8, rx: 3, fill: INK }, front));+ txt(front, 400, TOP + 66, "your orchestrator", { "text-anchor": "middle", "font-size": 27, "font-weight": 700 });+ el("circle", { cx: 150, cy: TOP + 58, r: 20, fill: "#fff", stroke: INK, "stroke-width": 3 }, front);+ const dial = el("path", { d: "M0 0 V-13", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round" }, front);+ el("rect", { x: 648, y: TOP + 32, width: 12, height: 58, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, front);+ const lever = el("rect", { x: 634, width: 40, height: 16, rx: 8, fill: YEL, stroke: INK, "stroke-width": 3 }, front);+ /* the vague note is still smoking a little */+ const puff = [0, 1].map(() => el("path", { d: "M0 0 q-9 -11 0 -22 t0 -22", fill: "none", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, svg));+ const cap = strip(svg, 508, 740);+ const PA = 3.3, PB = 7.6;+ return (t) => {+ const out = seg(t, 13.2, 13.45), vis = 1 - out, sink = 30 * io(seg(t, 13.2, 13.5));+ const cookA = t >= 2.05 && t < PA, cookB = t >= 6.35 && t < PB, cook = cookA || cookB;+ /* the slips: each pops out under its pad and drops into its slot */+ [[SA, XA, 0.9], [SB, XB, 5.2]].forEach(([g, x, a]) => {+ const f = io(seg(t, a + 0.45, a + 1.1));+ tf(g, x, (292 + (TOP + 70 - 292) * f).toFixed(2), land(seg(t, a, a + 0.3), 2).toFixed(3), (-8 * Math.sin(Math.PI * f)).toFixed(2));+ g.setAttribute("opacity", t >= a && t < a + 1.15 ? 1 : 0);+ });+ /* the toaster works, then pops */+ const lv = Math.max(seg(t, 1.95, 2.1) * (1 - seg(t, PA, PA + 0.06)), seg(t, 6.25, 6.4) * (1 - seg(t, PB, PB + 0.06)));+ const ha = t - PA, hb = t - PB, hp = (v) => (v > 0 && v < 1 ? 5 * Math.exp(-7 * v) * Math.abs(Math.sin(16 * v)) : 0);+ const jig = cook ? 1.3 * Math.sin(t * 60) : 0, lift = hp(ha) + hp(hb);+ [back, front].forEach((g) => g.setAttribute("transform", `translate(${jig.toFixed(2)} ${(-lift).toFixed(2)})`));+ lever.setAttribute("y", (TOP + 30 + 46 * lv).toFixed(2));+ slots[0].setAttribute("fill", cookA ? YEL : INK);+ slots[1].setAttribute("fill", cookB ? YEL : INK);+ const dk = cookA ? seg(t, 2.1, PA) : cookB ? seg(t, 6.4, PB) : 1;+ dial.setAttribute("transform", `translate(150 ${TOP + 58}) rotate(${(150 * (1 - dk)).toFixed(1)})`);+ /* the vague note comes back as one grey card with a question mark */+ const qa = land(seg(t, PA, PA + 0.42), 1.7), qw = t - PA - 0.4;+ tf(Q, XA, (TOP + 120 * (1 - qa) + sink).toFixed(2), 1, (qw > 0 ? 7 * Math.exp(-4 * qw) * Math.sin(13 * qw) : 0).toFixed(2));+ Q.setAttribute("opacity", t >= PA ? vis.toFixed(2) : 0);+ puff.forEach((c, j) => {+ const p0 = (t - PA - 0.1) * 0.4 - j * 0.5, p = p0 % 1;+ c.setAttribute("transform", `translate(${(XA + 64 + 8 * j + 4 * Math.sin(p * 5)).toFixed(1)} ${(TOP - 10 - 58 * p).toFixed(1)}) scale(${(0.7 + 0.45 * p).toFixed(3)})`);+ c.setAttribute("opacity", p0 < 0 ? 0 : (cl(p * 5) * (1 - p) * vis).toFixed(2));+ });+ /* the full brief comes back as three task cards that hop apart, and each gets its tick */+ const ta = land(seg(t, PB, PB + 0.4), 1.6), spr = io(seg(t, PB + 0.45, PB + 1.05));+ tasks.forEach((k, i) => {+ const hop = i === 1 ? 0 : 16 * Math.sin(Math.PI * spr);+ tf(k.g, (XB + (i - 1) * 86 * spr).toFixed(2), (TOP + 104 * (1 - ta) - hop + sink).toFixed(2), 1, ((i - 1) * 10 * Math.sin(Math.PI * spr)).toFixed(2));+ k.g.setAttribute("opacity", t >= PB ? vis.toFixed(2) : 0);+ const tk = seg(t, 9.1 + 0.45 * i, 9.4 + 0.45 * i);+ k.tick.setAttribute("stroke-dasharray", `${(21 * tk).toFixed(2)} 30`);+ k.tick.setAttribute("opacity", tk > 0 ? 1 : 0);+ k.box.setAttribute("fill", tk > 0.5 ? MINT : "#fff");+ });+ say(cap, t >= 13.2 || t < PA ? "A vague brief gets a vague job" : t < 5.2 ? "A vague one turns into a question" : t < PB ? "The outcome, the folder, and what done looks like" : "Turns into clear tasks", 700);+ };+ },+ };++ /* ---- team: three office chairs roll in, one per hire; a change goes from the builder to the reviewer and back with a tick ---- */+ SC.team = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Hire a small team", "tap the builder");+ const XS = [150, 400, 650], FL = 400, COL = [SKY, LIL, MINT], PR = [0.7, 2.6, 4.3], IN = [0.85, 2.75, 4.45];+ const ROLE = ["builder", "reviewer", "helper"], ENG = ["on Claude Code", "on Codex", "on a free engine"];+ el("path", { d: `M44 ${FL + 2} H756`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ /* three open seats, drawn as dotted outlines */+ const ghost = XS.map((x) => {+ const g = el("g", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round" }, svg);+ el("rect", { x: x - 50, y: FL - 214, width: 100, height: 92, rx: 22 }, g);+ el("rect", { x: x - 56, y: FL - 104, width: 112, height: 22, rx: 11 }, g);+ el("path", { d: `M${x} ${FL - 80} V${FL - 14} M${x - 38} ${FL - 6} H${x + 38}` }, g);+ return g;+ });+ /* under each seat: its role, and the engine it runs on once hired */+ const zone = XS.map((x, i) => {+ const g = el("g", {}, svg);+ if (i === 0) el("rect", { x: x - 112, y: 170, width: 224, height: 332, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const slot = el("rect", { x: x - 96, y: 458, width: 192, height: 36, rx: 18, fill: "none", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, g);+ const role = txt(g, x, 438, ROLE[i], { "text-anchor": "middle", "font-size": 26, "font-weight": 700, fill: MUTE });+ const p = pill(g, ENG[i], COL[i]);+ return { slot, role, p };+ });+ button(zone[0].role.parentNode, "Send another change from the builder to the reviewer", () => api.seek(6.5));+ /* the chairs */+ const chairs = COL.map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg)), top = el("g", {}, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, top);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, top);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, top);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, top);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, top);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, top);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, top);+ return { g, top };+ });+ /* the helper's papers, in a heap until they are tidied */+ const pile = [[-14, -16], [2, 9], [12, 22]].map(([dx, r]) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-9 -10 H9 M-9 0 H9 M-9 10 H3", stroke: MUTE, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, dx, r };+ });+ /* the change: a sheet that goes to the reviewer and comes back with a tick */+ const sheet = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -22, y: -28, width: 44, height: 56, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, sheet);+ const sl = [-12, 0, 12].map((y, j) => el("path", { d: `M-12 ${y} H${j === 2 ? 4 : 12}`, stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, sheet));+ const ok = el("g", {}, sheet);+ el("circle", { r: 12, fill: MINT, stroke: INK, "stroke-width": 2.5 }, ok);+ el("path", { d: "M-5.5 0 l4 4.5 l7.5 -9", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, ok);+ /* the button you press, and the pointer pressing it */+ const B = el("g", {}, svg);+ el("rect", { x: -100, y: -24, width: 200, height: 48, rx: 14, fill: YEL, stroke: INK, "stroke-width": 3 }, B);+ txt(B, 0, 8, "Add an agent", { "text-anchor": "middle", "font-size": 23, "font-weight": 700 });+ const rip = el("circle", { fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg);+ const cap = strip(svg, 508, 740);+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, idle = t > 11.8 && t < 13.2 ? 1 : 0;+ const press = Math.max(...PR.map((a) => bump(t, a)));+ tf(B, 400, 122, (1 - 0.06 * press + 0.012 * idle * Math.sin(now * 3)).toFixed(3));+ const lastP = PR.filter((a) => t >= a).pop(), rk = lastP === undefined ? 1 : seg(t, lastP, lastP + 0.6);+ rip.setAttribute("cx", 474); rip.setAttribute("cy", 138);+ rip.setAttribute("r", (8 + 30 * oc(rk)).toFixed(1)); rip.setAttribute("opacity", (rk < 1 ? 1 - rk : 0).toFixed(2));+ ptr.setAttribute("transform", `translate(${(474 + 2 * press).toFixed(2)} ${(138 + 3 * press + 2 * idle * Math.sin(now * 2)).toFixed(2)})`);+ /* each press rolls a chair in from the right; the third one arrives still spinning */+ chairs.forEach((ch, i) => {+ const k = seg(t, IN[i], IN[i] + 1.05), x = lerp(880, XS[i], land(k, 1.15)), st = t - IN[i] - 1.05;+ const lean = 3 * Math.sin(Math.PI * k) - (st > 0 ? 3 * Math.exp(-5 * st) * Math.sin(14 * st) : 0);+ tf(ch.g, x.toFixed(2), FL, 1, lean.toFixed(2));+ ch.g.setAttribute("opacity", t < IN[i] ? 0 : vis.toFixed(2));+ const sx = i === 2 ? Math.cos(4 * Math.PI * (1 - oc(seg(t, IN[2], IN[2] + 1.9)))) : 1;+ ch.top.setAttribute("transform", `scale(${sx.toFixed(3)} 1)`);+ ghost[i].setAttribute("opacity", Math.max(1 - seg(t, IN[i] + 0.6, IN[i] + 1.0), out).toFixed(2));+ const z = zone[i], on = t >= IN[i] + 1.0 && t < 13.2, pk = land(seg(t, IN[i] + 1.05, IN[i] + 1.4), 2.4) * vis;+ z.role.setAttribute("fill", on ? INK : MUTE);+ tf(z.p, XS[i], 476, pk.toFixed(3));+ z.slot.setAttribute("opacity", pk > 0.6 ? 0 : 1);+ });+ /* the change goes over, is read line by line, gets its tick and comes back */+ const g1 = io(seg(t, 7.2, 8.2)), g2 = io(seg(t, 9.9, 10.9)), back = t >= 9.9, k = back ? g2 : g1, lw = t - 10.9;+ const sx = lerp(XS[0], XS[1], back ? 1 - g2 : g1), sy = FL - 132 - 70 * Math.sin(Math.PI * k);+ const rot = (back ? -12 : 12) * Math.sin(Math.PI * k) + (lw > 0 ? 6 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0);+ tf(sheet, sx.toFixed(2), sy.toFixed(2), (land(seg(t, 6.6, 6.95), 2.2) * vis).toFixed(3), rot.toFixed(2));+ sl.forEach((l, j) => l.setAttribute("stroke", t > 8.45 + 0.3 * j && t < 13.2 ? INK : MUTE));+ tf(ok, 20, -24, land(seg(t, 9.45, 9.8), 2.6).toFixed(3));+ /* the helper's heap lands, then squares itself into a neat stack */+ pile.forEach((p, j) => {+ const dk = land(seg(t, 6.0 + 0.12 * j, 6.35 + 0.12 * j), 1.6), td = io(seg(t, 11.0 + 0.1 * j, 11.6 + 0.1 * j));+ tf(p.g, (XS[2] + lerp(p.dx, (j - 1) * 3, td)).toFixed(2), (FL - 130 - 60 * (1 - dk) - 3 * j * td).toFixed(2), 1, lerp(p.r, 0, td).toFixed(2));+ p.g.setAttribute("opacity", t < 6.0 + 0.12 * j ? 0 : vis.toFixed(2));+ });+ say(cap, t < PR[0] || t >= 13.2 ? "Press Add an agent" : t < PR[1] ? "A builder on Claude Code, writing the feature" : t < PR[2] ? "A reviewer on Codex, reading every change the builder makes" : t < 6.6 ? "A helper on a free engine, for docs and tidying" : t < 11 ? "Reading every change the builder makes" : "A first team of three", 700);+ };+ },+ };++ /* ---- schedule: press New automation, choose Schedule, fill in four things and walk off; the sun comes up and the card fires without you ---- */+ SC.schedule = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Work that starts without you", "tap the card");+ const X0 = 44, X1 = 500, Y0 = 88, Y1 = 474, HZ = 226, FL = 472, AX = 652, VX = 176, VW = 306, FY = [228, 288, 348, 420];+ const around = (n, cx, cy, s) => n.setAttribute("transform", `translate(${cx} ${cy}) scale(${s}) translate(${-cx} ${-cy})`);+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ const mini = (p, c) => {+ const g = el("g", {}, p);+ el("rect", { x: -9, y: -17, width: 18, height: 14, rx: 5, fill: c, stroke: INK, "stroke-width": 2 }, g);+ el("rect", { x: -11, y: -2, width: 22, height: 6, rx: 3, fill: c, stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M0 4 V11 M-8 13 H8", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ };+ /* the right side: the arc the sun climbs, the horizon, the floor and the agent's chair */+ el("path", { d: `M548 ${HZ} A104 104 0 0 1 756 ${HZ}`, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "2 10", "stroke-linecap": "round" }, svg);+ const cp = el("clipPath", { id: "mg-schedule-sky" }, el("defs", {}, svg));+ el("rect", { x: 500, y: 70, width: 300, height: HZ - 70 }, cp);+ const sun = mv(el("g", { "pointer-events": "none" }, el("g", { "clip-path": "url(#mg-schedule-sky)" }, svg)));+ const rays = el("g", {}, sun);+ for (let i = 0; i < 8; i++) el("path", { d: "M0 -35 V-46", stroke: INK, "stroke-width": 4, "stroke-linecap": "round", transform: `rotate(${i * 45})` }, rays);+ el("circle", { r: 25, fill: YEL, stroke: INK, "stroke-width": 3 }, sun);+ el("path", { d: `M524 ${HZ} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ el("path", { d: `M524 ${FL} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const ag = chair(svg, SKY, 0.78);+ /* the card: Automations, the New automation button, the two kinds, then the four fields */+ const card = el("g", {}, svg);+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 3 }, card);+ el("rect", { x: X0 - 9, y: Y0 - 9, width: X1 - X0 + 18, height: Y1 - Y0 + 18, rx: 24, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, card);+ button(card, "Fire the schedule again", () => api.seek(8.7));+ txt(card, 66, 128, "Automations", { "font-size": 24, "font-weight": 700 });+ const nb = el("g", {}, card);+ el("rect", { x: -98, y: -21, width: 196, height: 42, rx: 13, fill: YEL, stroke: INK, "stroke-width": 3 }, nb);+ txt(nb, 0, 7, "New automation", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ el("path", { d: `M${X0 + 2} 150 H${X1 - 2}`, stroke: SOFT, "stroke-width": 3 }, card);+ const k1 = pill(card, "Schedule", "#fff"), k2 = pill(card, "Webhook endpoint", "#fff");+ const w1 = parseFloat(k1.firstChild.getAttribute("width")), w2 = parseFloat(k2.firstChild.getAttribute("width")), KX = [66 + w1 / 2, 80 + w1 + w2 / 2];+ const names = ["Label", "When", "Sends to", "Message"].map((s, i) => txt(card, 66, FY[i] + 6, s, { "font-size": 18, "font-weight": 700, fill: MUTE }));+ const box = (y, h) => el("rect", { x: VX, y, width: VW, height: h, rx: 12, fill: "#fff", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, card);+ const bx = [box(FY[0] - 20, 40), box(FY[1] - 20, 40), box(FY[2] - 20, 40), box(FY[3] - 32, 66)];+ const wk = el("rect", { x: VX + VW / 2 + 4, y: FY[1] - 16, width: VW / 2 - 8, height: 32, rx: 9, fill: YEL, stroke: INK, "stroke-width": 2 }, card);+ el("path", { d: `M${VX + VW / 2} ${FY[1] - 13} V${FY[1] + 13}`, stroke: LIL, "stroke-width": 2.5 }, card);+ const segT = ["Interval", "Weekly"].map((s, i) => txt(card, VX + VW / 4 + (i * VW) / 2, FY[1] + 7, s, { "text-anchor": "middle", "font-size": 19, "font-weight": 600, fill: MUTE }));+ const vL = txt(card, VX + 16, FY[0] + 7, "", { "font-size": 19, "font-weight": 600 });+ const vM = [txt(card, VX + 16, FY[3] - 5, "", { "font-size": 19, "font-weight": 600 }), txt(card, VX + 16, FY[3] + 21, "", { "font-size": 19, "font-weight": 600 })];+ const caret = el("rect", { width: 2.5, height: 21, fill: INK }, card);+ const st = el("g", {}, card);+ mini(st, SKY).setAttribute("transform", "translate(14 2) scale(0.86)");+ txt(st, 34, 7, "an agent", { "font-size": 19, "font-weight": 700 });+ /* when it fires: a ripple off the card and a slip that flies to the chair */+ const rip = el("circle", { fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ const slip = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, slip);+ [-11, 0, 11].forEach((y, j) => el("path", { d: `M-10 ${y} H${j === 2 ? 3 : 10}`, stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, slip));+ const ok = el("g", {}, slip);+ el("circle", { r: 11, fill: MINT, stroke: INK, "stroke-width": 2.5 }, ok);+ el("path", { d: "M-5 0 l3.6 4 l7 -8.5", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, ok);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 508, 740);+ const PR = [0.85, 2.35, 3.25, 5.25, 5.95, 6.45];+ const KP = [[0, 572, 296], [0.6, 478, 130], [1.6, 478, 130], [2.15, 66 + w1 - 12, 188], [2.7, 66 + w1 - 12, 188], [3.1, 470, 236], [4.6, 470, 236], [5.05, 470, 296], [5.5, 470, 296], [5.8, 470, 356], [6.15, 470, 356], [6.4, 470, 436], [7.9, 470, 436], [8.6, 870, 420], [10.8, 870, 318], [11.3, 742, 318], [11.9, 742, 318], [12.35, 870, 318], [13.4, 870, 318], [14, 572, 296]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ const typed = (n, s, t, a, b) => { const c = Math.round(s.length * seg(t, a, b)), v = t < 13.2 ? s.slice(0, c) : s; if (n._v !== v) { n._v = v; n.textContent = v; } return t > a && t < b + 0.15; };+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, live = t < 13.2;+ const press = Math.max(...PR.map((a) => bump(t, a)));+ tf(nb, 392, 120, (1 - 0.07 * bump(t, PR[0])).toFixed(3));+ [k1, k2].forEach((k, i) => { tf(k, KX[i], 182, land(seg(t, 1.05 + 0.12 * i, 1.4 + 0.12 * i), 2).toFixed(3)); k.setAttribute("opacity", vis.toFixed(2)); });+ k1.firstChild.setAttribute("fill", t >= PR[1] + 0.12 && live ? LIL : "#fff");+ /* the fields wake up once Schedule is chosen */+ bx.forEach((b, i) => { const on = t >= 2.6 + 0.12 * i && live; b.setAttribute("stroke", on ? INK : LIL); b.setAttribute("stroke-dasharray", on ? "none" : "6 7"); names[i].setAttribute("fill", on ? INK : MUTE); });+ const wkOn = t >= PR[3] + 0.1 && live;+ around(wk, VX + (3 * VW) / 4, FY[1], land(seg(t, PR[3] + 0.05, PR[3] + 0.35), 2.4).toFixed(3));+ wk.setAttribute("opacity", vis.toFixed(2));+ segT[1].setAttribute("fill", wkOn ? INK : MUTE);+ tf(st, VX + 8, FY[2], land(seg(t, PR[4] + 0.05, PR[4] + 0.4), 2.4).toFixed(3));+ st.setAttribute("opacity", vis.toFixed(2));+ [vL, ...vM].forEach((n) => n.setAttribute("opacity", vis.toFixed(2)));+ const tL = typed(vL, "morning check", t, 3.4, 4.4), t1 = typed(vM[0], "check something and", t, 6.6, 7.25), t2 = typed(vM[1], "write you three lines", t, 7.3, 7.95);+ const cn = tL ? vL : t1 ? vM[0] : t2 ? vM[1] : null;+ caret.setAttribute("opacity", cn && Math.sin(now * 9) > -0.3 ? 1 : 0);+ if (cn) { caret.setAttribute("x", (parseFloat(cn.getAttribute("x")) + cn.getComputedTextLength() + 3).toFixed(1)); caret.setAttribute("y", parseFloat(cn.getAttribute("y")) - 17); }+ /* the sun climbs, the card fires, the slip flies to the chair and is ticked */+ const rise = io(seg(t, 8.6, 10.0)) * (1 - io(seg(t, 13.0, 13.6))), a = ((212 - 74 * rise) * Math.PI) / 180;+ tf(sun, (AX + 104 * Math.cos(a)).toFixed(2), (HZ - 104 * Math.sin(a)).toFixed(2));+ rays.setAttribute("transform", `rotate(${((now * 14) % 360).toFixed(1)})`);+ const pulse = bump(t, 10.0);+ around(card, (X0 + X1) / 2, (Y0 + Y1) / 2, (1 + 0.014 * pulse).toFixed(4));+ const rk = seg(t, 10.0, 10.6);+ rip.setAttribute("cx", X1); rip.setAttribute("cy", 418);+ rip.setAttribute("r", (8 + 40 * oc(rk)).toFixed(1)); rip.setAttribute("opacity", (rk > 0 && rk < 1 ? 1 - rk : 0).toFixed(2));+ const fk = io(seg(t, 10.4, 11.2)), lw = t - 11.2;+ const sx = lerp(530, AX, fk), sy = lerp(418, FL - 81 - 24, fk) - 70 * Math.sin(Math.PI * fk);+ const rot = -12 * Math.sin(Math.PI * fk) + (lw > 0 ? 7 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0);+ tf(slip, sx.toFixed(2), sy.toFixed(2), land(seg(t, 10.05, 10.35), 2.2).toFixed(3), rot.toFixed(2));+ slip.setAttribute("opacity", t < 10.05 ? 0 : vis.toFixed(2));+ tf(ok, 16, -20, land(seg(t, 11.35, 11.7), 2.6).toFixed(3));+ const hop = 8 * Math.sin(Math.PI * seg(t, 6.0, 6.35)) + 6 * Math.sin(Math.PI * seg(t, 11.15, 11.45));+ tf(ag, AX, (FL - hop).toFixed(2));+ /* the pointer: fills the card, walks off, and peeks back in once to check */+ const [px, py] = at(t), wig = t > 11.3 && t < 11.9 ? 3 * Math.sin((t - 11.3) * 26) * Math.sin((Math.PI * (t - 11.3)) / 0.6) : 0;+ ptr.setAttribute("transform", `translate(${(px + 2 * press + wig).toFixed(2)} ${(py + 3 * press).toFixed(2)})`);+ say(cap, t < 0.9 || t >= 13.2 ? "Open Automations and press New automation" : t < 2.45 ? "Choose Schedule" : t < 4.9 ? "Then fill in four things" : t < 6.3 ? "When: an Interval or Weekly" : t < 8.0 ? "The message is what the agent should do when it fires" : t < 10.0 ? "A schedule sends a message to an agent on a clock" : "Every weekday morning, ask an agent to check something", 700);+ };+ },+ };++ /* ---- slack: switch Slack on under Connections and pick who answers; each card sends a note down its own string and the reply comes back into the same thread. Once, the strings go slack ---- */+ SC.slack = {+ dur: 14, still: 12.6,+ build(svg, api) {+ head(svg, "Send work from a channel", "tap a card");+ const CY = [176, 290, 404], XC = 628, FL = 474, S = 0.72, E = [XC - 46, FL - 80], M = [4.6, 5.4, 6.2], IN = 2.45;+ const NAMES = ["a channel", "another channel", "a direct message"];+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ /* Settings, then Connections: the Slack switch and Every channel the bot is in */+ el("rect", { x: 44, y: 84, width: 712, height: 72, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 66, 111, "Settings", { "font-size": 16, "font-weight": 600, fill: MUTE });+ txt(svg, 66, 139, "Connections", { "font-size": 22, "font-weight": 700 });+ txt(svg, 248, 128, "Slack", { "font-size": 20, "font-weight": 700 });+ const trk = el("rect", { x: 314, y: 106, width: 54, height: 28, rx: 14, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ const knob = el("circle", { cy: 120, r: 9, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const cb = el("rect", { x: 398, y: 107, width: 26, height: 26, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const tick = el("path", { d: "M404.5 120.5 l5 5.5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round", pathLength: 1, "stroke-dasharray": "0 1" }, svg);+ fit(txt(svg, 436, 128, "Every channel the bot is in", { "font-size": 20, "font-weight": 600 }), 300);+ /* who answers: an empty seat until the chair rolls in */+ fit(txt(svg, XC - 12, 292, "Who answers inbound messages", { "text-anchor": "middle", "font-size": 18, "font-weight": 700 }), 290);+ el("path", { d: `M440 ${FL + 2} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const ghost = el("g", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round", transform: `translate(${XC} ${FL}) scale(${S})` }, svg);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22 }, ghost);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11 }, ghost);+ el("path", { d: "M0 -80 V-14 M-38 -6 H38" }, ghost);+ /* one string from each card to the chair */+ const P0 = CY.map((y) => [372, y + 50]);+ const strs = CY.map(() => el("path", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", pathLength: 1 }, svg));+ const ctl = (i, sag) => [(P0[i][0] + E[0]) / 2, (P0[i][1] + E[1]) / 2 + sag];+ const bez = (i, s, sag) => { const [x0, y0] = P0[i], [cx, cy] = ctl(i, sag), u = 1 - s; return [u * u * x0 + 2 * u * s * cx + s * s * E[0], u * u * y0 + 2 * u * s * cy + s * s * E[1]]; };+ /* the three chat cards: a channel, another channel, a direct message */+ const cards = CY.map((y0, i) => {+ const g = el("g", {}, svg);+ el("rect", { x: 44, y: y0, width: 328, height: 100, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: 37, y: y0 - 7, width: 342, height: 114, rx: 21, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ if (i < 2) el("path", { d: `M58 ${y0 + 15} h16 a5 5 0 0 1 5 5 v7 a5 5 0 0 1 -5 5 h-9 l-6 5 v-5 h-1 a5 5 0 0 1 -5 -5 v-7 a5 5 0 0 1 5 -5 z`, fill: SKY, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, g);+ else el("circle", { cx: 66, cy: y0 + 24, r: 9, fill: YEL, stroke: INK, "stroke-width": 2 }, g);+ txt(g, 88, y0 + 31, NAMES[i], { "font-size": 20, "font-weight": 700 });+ const msg = el("g", {}, g);+ el("rect", { x: 0, y: -11, width: 200, height: 22, rx: 11, fill: SOFT, stroke: INK, "stroke-width": 2 }, msg);+ el("path", { d: "M14 0 H122 M134 0 H180", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, msg);+ const rep = el("g", {}, g);+ el("path", { d: "M10 13 V22 Q10 30 18 30 H25", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, rep);+ const av = el("g", { transform: "translate(40 32) scale(0.74)" }, rep);+ el("rect", { x: -9, y: -17, width: 18, height: 14, rx: 5, fill: SKY, stroke: INK, "stroke-width": 2.4 }, av);+ el("rect", { x: -11, y: -2, width: 22, height: 6, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2.4 }, av);+ el("path", { d: "M0 4 V11 M-8 13 H8", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, av);+ el("rect", { x: 56, y: 20, width: 200, height: 21, rx: 10.5, fill: MINT, stroke: INK, "stroke-width": 2 }, rep);+ el("path", { d: "M70 30.5 H170 M182 30.5 H232", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, rep);+ button(g, `Send a message from ${NAMES[i]}`, () => api.seek(M[i] - 0.15));+ return { msg, rep, y0 };+ });+ const ag = chair(svg, SKY, S);+ const note = CY.map(() => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -11, y: -14, width: 22, height: 28, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-5 -5 H5 M-5 2 H5 M-5 9 H1", stroke: MUTE, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ });+ const back = CY.map(() => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -17, y: -10, width: 34, height: 20, rx: 10, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-8 0 H8", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ });+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ const PR = [0.6, 1.45, 2.3];+ const KP = [[0, 540, 232], [0.45, 354, 126], [0.95, 354, 126], [1.25, 412, 128], [1.75, 412, 128], [2.1, 652, 402], [2.45, 652, 402], [3.0, 880, 232], [13.4, 880, 232], [14, 540, 232]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, live = t < 13.2;+ /* the switch: on at the first press; once it drops out and comes back by itself */+ const g1 = io(seg(t, 10.4, 10.6)), g2 = land(seg(t, 11.3, 11.6), 2.4), on = io(seg(t, PR[0], PR[0] + 0.25)) * (1 - io(seg(t, 13.2, 13.6))) * (1 - g1 + g1 * g2);+ knob.setAttribute("cx", (328 + 26 * on).toFixed(2));+ trk.setAttribute("fill", on > 0.5 ? MINT : SOFT);+ const chk = t >= PR[1] + 0.05 && live;+ cb.setAttribute("fill", chk ? YEL : "#fff");+ tick.setAttribute("stroke-dasharray", `${(oc(seg(t, PR[1] + 0.05, PR[1] + 0.35)) * (live ? 1 : 0)).toFixed(3)} 1`);+ /* the chair rolls in to the empty seat */+ const k = seg(t, IN, IN + 1.0), x = lerp(880, XC, land(k, 1.15)), st = t - IN - 1.0;+ const lean = 3 * Math.sin(Math.PI * k) - (st > 0 ? 3 * Math.exp(-5 * st) * Math.sin(14 * st) : 0);+ const hop = M.reduce((h, m) => h + 6 * Math.sin(Math.PI * seg(t, m + 1.05, m + 1.35)), 0);+ tf(ag, x.toFixed(2), (FL - hop).toFixed(2), 1, lean.toFixed(2));+ ag.setAttribute("opacity", t < IN ? 0 : vis.toFixed(2));+ ghost.setAttribute("opacity", Math.max(1 - seg(t, IN + 0.6, IN + 1.0), out).toFixed(2));+ /* the strings: drawn out, taut; slack while Slack is off; a twang when it comes back */+ const tau = t - 11.3, twang = tau > 0 && tau < 1.2 ? 9 * Math.exp(-4 * tau) * Math.sin(22 * tau) : 0;+ const slack = 66 * io(seg(t, 10.4, 10.8)) * (1 - oc(seg(t, 11.3, 11.75))) + 3 * Math.sin(now * 3) * seg(t, 10.6, 10.9) * (1 - seg(t, 11.3, 11.5));+ const sag = 8 + slack + twang + (t > 12.2 && live ? 1.5 * Math.sin(now * 2) : 0);+ strs.forEach((p, i) => {+ const [cx, cy] = ctl(i, sag);+ p.setAttribute("d", `M${P0[i][0]} ${P0[i][1]} Q${cx.toFixed(1)} ${cy.toFixed(1)} ${E[0]} ${E[1]}`);+ p.setAttribute("stroke-dasharray", `${io(seg(t, 3.5 + 0.12 * i, 4.2 + 0.12 * i)).toFixed(3)} 1`);+ p.setAttribute("opacity", vis.toFixed(2));+ });+ /* each card: its message, the note down the string, the reply back, the reply in the thread */+ cards.forEach((c, i) => {+ const m = M[i];+ tf(c.msg, 66, c.y0 + 50, (land(seg(t, m, m + 0.35), 2.2) * vis).toFixed(3));+ tf(c.rep, 66, c.y0 + 50, (land(seg(t, m + 2.2, m + 2.55), 2.2) * vis).toFixed(3));+ const nk = io(seg(t, m + 0.25, m + 1.05)), [nx, ny] = bez(i, 0.07 + 0.89 * nk, sag);+ tf(note[i], nx.toFixed(2), ny.toFixed(2), land(seg(t, m + 0.25, m + 0.45), 2).toFixed(3), (8 * Math.sin(Math.PI * nk)).toFixed(2));+ note[i].setAttribute("opacity", t > m + 0.25 && t < m + 1.1 ? 1 : 0);+ const rk = io(seg(t, m + 1.4, m + 2.2)), [rx, ry] = bez(i, 0.96 - 0.88 * rk, sag);+ tf(back[i], rx.toFixed(2), ry.toFixed(2), land(seg(t, m + 1.4, m + 1.6), 2).toFixed(3), (-8 * Math.sin(Math.PI * rk)).toFixed(2));+ back[i].setAttribute("opacity", t > m + 1.4 && t < m + 2.22 ? 1 : 0);+ });+ const press = Math.max(...PR.map((a) => bump(t, a))), [px, py] = at(t);+ ptr.setAttribute("transform", `translate(${(px + 2 * press).toFixed(2)} ${(py + 3 * press).toFixed(2)})`);+ say(cap, t < 1.0 || t >= 13.2 ? "Under Settings, then Connections" : t < 2.0 ? "One bot can listen to Every channel the bot is in" : t < 3.6 ? "You choose the agent under Who answers inbound messages" : t < 6.2 ? "Send work from a channel and read the answer in the same thread" : t < 9.0 ? "A direct message to the bot is always answered" : t < 10.4 ? "The reply comes back in the same thread" : t < 12.0 ? "If Slack fails to start, it now comes back by itself" : "Read the answer in the same thread", 700);+ };+ },+ };++ /* ---- temp: a message comes in, a folding chair is opened (at the second try) beside the lasting chairs, takes the one job, reports done, folds flat and leaves ---- */+ SC.temp = {+ dur: 14, still: 9.9,+ build(svg, api) {+ head(svg, "A temp is an agent for one job", "tap the message");+ const FL = 444, XS = [124, 280, 436], COL = [SKY, LIL, MINT], TX = 640, FS = 1.15, HOME = [124, 140], WAIT = [612, 150], SEAT = [612, 318];+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ /* a folding chair seen from the side: back post, seat and front leg turn on one bolt */+ const folding = (p, ghost) => {+ const g = el("g", ghost ? { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round" } : { "pointer-events": "none" }, p);+ const ln = ghost ? {} : { stroke: INK, "stroke-width": 7.5, "stroke-linecap": "round" }, sh = ghost ? {} : { fill: YEL, stroke: INK, "stroke-width": 3 };+ const leg = el("path", Object.assign({ d: "M24 -92 L-40 0" }, ln), g);+ el("path", Object.assign({ d: "M24 0 V-176" }, ln), g);+ const seat = el("rect", Object.assign({ x: -58, y: -92, width: 88, height: 18, rx: 8 }, sh), g);+ el("rect", Object.assign({ x: 6, y: -188, width: 38, height: 60, rx: 10 }, sh), g);+ if (!ghost) el("circle", { cx: 24, cy: -92, r: 4.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ return { g, leg, seat };+ };+ el("path", { d: `M44 ${FL + 2} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const gh = folding(svg, true);+ tf(gh.g, TX, FL, FS);+ /* the lasting agents: the three office chairs from the team */+ const lc = XS.map((x, i) => chair(svg, COL[i], 0.95));+ el("path", { d: "M58 462 V472 H502 V462", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ txt(svg, 280, 500, "your lasting agents", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ /* the tag that hangs on the temp while it is here */+ const tag = el("g", {}, svg);+ const tp = pill(tag, "A temp", YEL), tw = parseFloat(tp.firstChild.getAttribute("width"));+ tf(tp, 716, 192);+ el("path", { d: `M${TX + 30} 232 Q${TX + 34} 206 ${716 - tw / 2 + 8} 196`, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, tag);+ const fw = el("g", {}, svg), fc = folding(fw, false);+ /* the slot the message waits in, which is also the tap target */+ const slot = el("g", {}, svg);+ el("rect", { x: HOME[0] - 72, y: HOME[1] - 36, width: 144, height: 72, rx: 18, fill: "#fff", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, slot);+ el("rect", { x: HOME[0] - 80, y: HOME[1] - 44, width: 160, height: 92, rx: 22, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, slot);+ button(slot, "Send another message to a temp", () => api.seek(0.7));+ const dots = [0, 1, 2].map((j) => el("circle", { cx: SEAT[0] - 30 + 18 * j, cy: 270, r: 5.5, fill: LIL, stroke: INK, "stroke-width": 2 }, svg));+ const msg = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M-32 24 L-42 42 L-14 24 Z", fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, msg);+ el("rect", { x: -62, y: -28, width: 124, height: 56, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, msg);+ txt(msg, 0, 8, "one job", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ const done = mv(el("g", { "pointer-events": "none" }, svg));+ pill(done, "done", MINT, true);+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out;+ /* the folding chair: slides in folded, half opens, snaps shut, opens properly; later folds and is dragged off */+ let u = 0;+ if (t >= 2.3 && t < 2.65) u = 0.5 * io(seg(t, 2.3, 2.65));+ else if (t >= 2.65 && t < 3.15) u = 0.5 - 0.42 * oc(seg(t, 2.65, 2.85));+ else if (t >= 3.15 && t < 8.1) u = 0.08 + 0.92 * land(seg(t, 3.15, 3.65), 1.4);+ else if (t >= 8.1 && t < 8.6) u = 1 - io(seg(t, 8.1, 8.5));+ const inK = seg(t, 1.1, 2.1), outK = io(seg(t, 9.0, 11.6));+ const fx = t < 8.6 ? lerp(900, TX, land(inK, 1.0)) : lerp(TX, 930, outK);+ const sh = t > 2.85 && t < 3.15 ? 1.6 * Math.sin((t - 2.85) * 50) * (1 - seg(t, 2.85, 3.15)) : 0;+ const lean = -8 * Math.sin(Math.PI * inK) + 12 * io(seg(t, 8.6, 9.0)) + sh;+ fc.leg.setAttribute("transform", `rotate(${(-34.8 * (1 - u)).toFixed(2)} 24 -92)`);+ fc.seat.setAttribute("transform", `rotate(${(90 * (1 - u)).toFixed(2)} 24 -92)`);+ tf(fc.g, fx.toFixed(2), (FL - 20 * FS * cl(1 - u)).toFixed(2), FS, lean.toFixed(2));+ gh.g.setAttribute("opacity", (1 - cl(u)).toFixed(2));+ /* marked as moving only while it slides in or out */+ if ((t > 1.1 && t < 2.3) || t > 8.6) fw.setAttribute("data-mv", ""); else fw.removeAttribute("data-mv");+ const tk = land(seg(t, 3.7, 4.0), 2.2) * (1 - seg(t, 7.9, 8.2));+ tag.setAttribute("opacity", tk > 0.02 ? Math.min(1, tk).toFixed(2) : 0);+ tf(tp, 716, 192, Math.min(1.2, tk).toFixed(3));+ /* the message: waits, flies over, hovers through the struggle, lands, is worked, and a new one arrives */+ let mx = HOME[0], my = HOME[1] + 3 * Math.sin(now * 2.2), ms = 1, mr = 0, mo = 1;+ if (t >= 0.8 && t < 1.0) ms = 1 + 0.06 * Math.sin(Math.PI * seg(t, 0.8, 1.0));+ else if (t >= 1.0 && t < 2.2) { const k = io(seg(t, 1.0, 2.2)); mx = lerp(HOME[0], WAIT[0], k); my = lerp(HOME[1], WAIT[1], k) - 22 * Math.sin(Math.PI * k); mr = 6 * Math.sin(Math.PI * k); }+ else if (t >= 2.2 && t < 3.85) { mx = WAIT[0]; my = WAIT[1] + 4 * Math.sin((t - 2.2) * 6); }+ else if (t >= 3.85 && t < 9.0) { const k = land(seg(t, 3.85, 4.35), 1.3); mx = lerp(WAIT[0], SEAT[0], cl(k)); my = lerp(WAIT[1], SEAT[1], k); ms = lerp(1, 0.8, cl(k)) * (1 - io(seg(t, 6.2, 6.5))); mo = t < 6.5 ? 1 : 0; }+ else if (t >= 9.0) { const k = land(seg(t, 9.0, 9.6), 1.4); mx = lerp(-90, HOME[0], k); mo = seg(t, 9.0, 9.12); }+ tf(msg, mx.toFixed(2), my.toFixed(2), ms.toFixed(3), mr.toFixed(2));+ msg.setAttribute("opacity", mo);+ dots.forEach((d, j) => { const w = t > 4.45 && t < 5.65 ? 1 : 0; d.setAttribute("opacity", w); d.setAttribute("cy", (270 - w * 6 * Math.max(0, Math.sin((t - 4.45) * 7 - j * 0.9))).toFixed(2)); });+ /* done pops up, flies to the first lasting chair, and that chair gives a small hop */+ const dk = io(seg(t, 6.7, 7.6)), lw = t - 7.6;+ const dx = lerp(SEAT[0] - 18, XS[0], dk), dy = lerp(222, 326, dk) - 90 * Math.sin(Math.PI * dk);+ tf(done, dx.toFixed(2), dy.toFixed(2), (land(seg(t, 5.7, 6.0), 2.4) * vis).toFixed(3), ((lw > 0 ? 6 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0) - 10 * Math.sin(Math.PI * dk)).toFixed(2));+ done.setAttribute("opacity", t < 5.7 ? 0 : vis.toFixed(2));+ lc.forEach((g, i) => tf(g, XS[i], (FL - (i === 0 ? 7 * Math.sin(Math.PI * seg(t, 7.6, 7.9)) : 0)).toFixed(2)));+ say(cap, t < 0.8 || t >= 13.2 ? "There is no button for one. You ask" : t < 3.8 ? "A temp can answer inbound messages" : t < 5.7 ? "It gets an objective, does the work" : t < 10.6 ? "Reports done and leaves" : "A temp answers one message or does one job", 700);+ };+ },+ };++ /* ---- tokens: slips fill a tall glass to the line and a press squashes them into one summary; on a switch of work the+ glass is turned over and shaken out (the summary clings on for one more shake), and a handoff.md note lands in it ---- */+ SC.tokens = {+ dur: 14, still: 12.6,+ build(svg, api) {+ head(svg, "Keep sessions small", "tap the glass");+ const GX = 214, GY = 356, PX = 360, PY = 196;+ /* the two rules, in Automations under Context, both switched on */+ el("rect", { x: 446, y: 92, width: 310, height: 290, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 468, 126, "Automations", { "font-size": 17, "font-weight": 600, fill: MUTE });+ txt(svg, 468, 158, "Context", { "font-size": 26, "font-weight": 700 });+ const lab = [];+ const rows = [["Compact when", "context fills"], ["Clear on a switch", "of work"]].map((L, i) => {+ const y = 176 + i * 102, r = el("rect", { x: 462, y, width: 278, height: 90, rx: 14, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ L.forEach((s, j) => lab.push(fit(txt(svg, 480, y + 38 + j * 30, s, { "font-size": 21, "font-weight": 700 }), 184)));+ el("rect", { x: 676, y: y + 32, width: 50, height: 26, rx: 13, fill: MINT, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 713, cy: y + 45, r: 8.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ return r;+ });+ const fs = Math.min(...lab.map((n) => parseFloat(n.getAttribute("font-size"))));+ lab.forEach((n) => n.setAttribute("font-size", fs));+ const keep = el("g", {}, svg);+ pill(keep, "Leave them on", MINT, true);+ /* the coaster and the glass; the slips, the summary and the note live inside the glass and turn with it */+ el("rect", { x: GX - 110, y: 476, width: 220, height: 10, rx: 5, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ const G = mv(el("g", {}, svg));+ const OUT = "M-88 -110 L-76 108 Q-75 118 -65 118 H65 Q75 118 76 108 L88 -110";+ el("path", { d: OUT + " Z", fill: "#fff" }, G);+ const C = el("g", {}, G);+ el("path", { d: "M-76 100 H76 L76 108 Q75 118 65 118 H-65 Q-75 118 -76 108 Z", fill: SOFT }, G);+ el("path", { d: "M-80 -58 H80", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "8 8", "stroke-linecap": "round" }, G);+ el("path", { d: "M-64 -88 L-58 40", stroke: SKY, "stroke-width": 7, "stroke-linecap": "round" }, G);+ el("path", { d: OUT, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round", "stroke-linecap": "round" }, G);+ el("path", { d: "M-76 100 H76", stroke: INK, "stroke-width": 2 }, G);+ const REST = [];+ for (let i = 0; i < 14; i++) REST.push([i % 2 ? 34 : -34, 89 - Math.floor(i / 2) * 22]);+ REST.push([-34, 53], [34, 53], [-28, 31]);+ const ROT = [-5, 4, 3, -4, -3, 5, 4, -3, -5, 2, 3, -4, -2, 4, 3, -4, 6];+ const TL = REST.map((_, i) => (i < 14 ? 0.7 + i * 0.22 : 6.8 + (i - 14) * 0.3));+ const S = REST.map(() => {+ const g = el("g", { "pointer-events": "none" }, C);+ el("rect", { x: -31, y: -9, width: 62, height: 18, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-21 0 H13", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ });+ const sum = el("g", {}, C);+ el("rect", { x: -66, y: -17, width: 132, height: 34, rx: 6, fill: MINT, stroke: INK, "stroke-width": 2.5 }, sum);+ const sumT = txt(sum, 0, 6, "summary", { "text-anchor": "middle", "font-size": 18, "font-weight": 700 });+ const ho = el("g", {}, C);+ el("rect", { x: -66, y: -19, width: 132, height: 38, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, ho);+ txt(ho, 0, 6, "handoff.md", { "text-anchor": "middle", "font-size": 18, "font-weight": 700, "font-family": MONO });+ /* the press: a knob, a rod and a mesh disc, parked beside the glass */+ const P = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M0 -6 V-54", stroke: INK, "stroke-width": 6, "stroke-linecap": "round" }, P);+ el("circle", { cx: 0, cy: -66, r: 13, fill: YEL, stroke: INK, "stroke-width": 2.5 }, P);+ el("rect", { x: -70, y: -6, width: 140, height: 12, rx: 6, fill: SKY, stroke: INK, "stroke-width": 2.5 }, P);+ el("path", { d: "M-46 -6 V6 M-23 -6 V6 M0 -6 V6 M23 -6 V6 M46 -6 V6", stroke: INK, "stroke-width": 1.5 }, P);+ const hit = el("g", {}, svg);+ el("rect", { x: GX - 106, y: 230, width: 212, height: 266, rx: 18, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Fill the glass again", () => api.seek(0.3));+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the glass: turned over, shaken twice, turned back */+ const flip = io(seg(t, 8.0, 8.6)) + io(seg(t, 10.5, 11.1));+ const lift = 24 * (io(seg(t, 8.0, 8.4)) - io(seg(t, 10.7, 11.1)));+ const shA = t > 8.6 && t < 9.2 ? Math.sin((t - 8.6) * 46) * (1 - seg(t, 8.6, 9.2)) : 0;+ const shB = t > 9.5 && t < 10.1 ? Math.sin((t - 9.5) * 50) * (1 - seg(t, 9.5, 10.1)) : 0;+ const st = t - 11.1, settle = st > 0 && st < 0.7 ? 3 * Math.exp(-6 * st) * Math.sin(18 * st) : 0;+ const gr = 180 * flip + 5 * shA + 8 * shB + settle;+ tf(G, (GX + 7 * shA + 12 * shB).toFixed(2), (GY - lift).toFixed(2), 1, gr.toFixed(2));+ /* the press: parked with a small bob, swings over, presses down, comes back */+ const bob = (1 - seg(t, 3.8, 4.1) + seg(t, 6.8, 7.1)) * 2 * Math.sin(now * 2.2);+ let px = PX, py = PY + bob;+ if (t >= 4.1 && t < 4.8) { const k = io(seg(t, 4.1, 4.8)); px = lerp(PX, GX, k); py = lerp(PY, GY - 122, k) - 34 * Math.sin(Math.PI * k); }+ else if (t >= 4.8 && t < 5.65) { px = GX; py = GY + lerp(-122, 56, io(seg(t, 4.8, 5.4))); }+ else if (t >= 5.65 && t < 6.05) { px = GX; py = GY + lerp(56, -122, oc(seg(t, 5.65, 6.05))); }+ else if (t >= 6.05 && t < 6.8) { const k = seg(t, 6.05, 6.8); px = lerp(GX, PX, io(k)); py = lerp(GY - 122, PY, land(k, 1.2)) - 30 * Math.sin(Math.PI * io(k)); }+ tf(P, px.toFixed(2), py.toFixed(2));+ /* the slips: drop in, pile up, get squashed; the later ones fall out when the glass is turned over */+ const d = t >= 4.8 && t < 6.05 ? py - GY : -999, sq = cl((100 - (d + 6)) / 152, 0.12, 1), gone = seg(t, 5.42, 5.6);+ S.forEach((g, i) => {+ const [rx, ry] = REST[i], a = TL[i] - 0.45;+ let x = rx, y = ry, r = ROT[i], s = 1, sy = 1, o = t < a ? 0 : 1;+ if (t < TL[i]) { const k = seg(t, a, TL[i]); x = lerp(rx * 0.6, rx, io(k)); y = lerp(-196, ry, k * k); r = ROT[i] * (5 - 4 * k); s = Math.min(1.2, land(seg(t, a, a + 0.18), 2)); }+ else { const b = t - TL[i]; if (b < 0.3) y -= 6 * Math.sin((Math.PI * b) / 0.3) * (1 - b / 0.3); }+ if (i < 14) { y = 100 - (100 - y) * sq; sy = Math.min(1, Math.max(0.3, sq * 1.15)); o *= 1 - gone; }+ else if (t > 8.6) { const f0 = 8.62 + (i - 14) * 0.07, u = seg(t, f0, f0 + 0.45); y = lerp(ry, -180, u * u); r += 25 * u * (i % 2 ? 1 : -1); o *= 1 - seg(u, 0.55, 0.95); }+ g.setAttribute("transform", `translate(${x.toFixed(2)} ${y.toFixed(2)}) rotate(${r.toFixed(2)}) scale(${s.toFixed(3)} ${(s * sy).toFixed(3)})`);+ g.setAttribute("opacity", o.toFixed(2));+ g.setAttribute("display", o < 0.01 ? "none" : "inline");+ });+ /* the summary: pops out of the squash, clings through the first shake, drops on the second */+ let sY = 81, sR = 0, so = t < 5.4 ? 0 : 1;+ if (t > 8.6 && t < 9.75) sY = 81 + 1.5 * shA;+ if (t >= 9.75) { const u = seg(t, 9.75, 10.25); sY = lerp(81, -180, u * u); sR = -30 * u; so = 1 - seg(u, 0.55, 0.95); }+ tf(sum, 0, sY.toFixed(2), Math.min(1.3, land(seg(t, 5.4, 5.7), 2.2)).toFixed(3), sR.toFixed(2));+ /* the slip turns with the glass; its word flips over halfway, while it is on its side, so it never reads upside down */+ sumT.setAttribute("transform", Math.round(flip) === 1 ? "rotate(180 0 -0.5)" : "");+ sum.setAttribute("opacity", so.toFixed(2));+ sum.setAttribute("display", so < 0.01 ? "none" : "inline");+ /* the handoff note drops into the clean glass and stays */+ const hk = seg(t, 11.25, 11.7), hb = t - 11.7;+ const hy = t < 11.7 ? lerp(-196, 79, hk * hk) : 79 - (hb < 0.35 ? 7 * Math.sin((Math.PI * hb) / 0.35) * (1 - hb / 0.35) : 0);+ tf(ho, 0, hy.toFixed(2), Math.min(1.2, land(seg(t, 11.25, 11.45), 2)).toFixed(3), (t < 11.7 ? 8 * (1 - hk) : 0).toFixed(2));+ const hoo = t < 11.25 ? 0 : 1 - out;+ ho.setAttribute("opacity", hoo.toFixed(2));+ ho.setAttribute("display", hoo < 0.01 ? "none" : "inline");+ rows[0].setAttribute("fill", t >= 3.9 && t < 6.4 ? YEL : SOFT);+ rows[1].setAttribute("fill", t >= 7.7 && t < 11.9 ? YEL : SOFT);+ tf(keep, 601, 432, (1 + 0.12 * Math.sin(Math.PI * seg(t, 12.0, 12.35))).toFixed(3));+ say(cap, t < 0.5 || t >= 13.3 ? "Two defaults in 0.5.6 keep sessions small" : t < 3.9 ? "Long sessions fill an agent's context" : t < 6.4 ? "Compact when context fills summarises a long session" : t < 7.7 ? "and carries the summary forward" : t < 11.2 ? "Clear on a switch of work clears the session" : t < 12.0 ? "and leaves a handoff file to carry the thread" : "Both rules are in Automations under Context", 700);+ };+ },+ };++ /* ---- sidebar: the Pro view's six rows; a highlight walks down them and each row writes its one line in the pane.+ Once it is picked, the Stapler row floats a little above the sidebar ---- */+ SC.sidebar = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "One window with a sidebar", "tap a row");+ const X0 = 44, X1 = 756, Y0 = 88, YB = 140, Y1 = 498, SW = 262, NX = 304;+ const RY = [182, 238, 294, 350, 406, 462], LIT = [1.0, 2.7, 4.4, 6.1, 7.8, 9.5];+ const NAMES = ["Tasks", "Inbox", "Automations", "Memory", "Capabilities", "Stapler"];+ const NOTE = ["every card in the office", "one conversation per agent", "Scheduled, Context and Webhooks", "Search every agent's memory", "what each agent can reach", "a small window that floats above every app"];+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "#fff" }, svg);+ el("path", { d: `M${X0} ${YB} H${SW} V${Y1} H${X0 + 18} Q${X0} ${Y1} ${X0} ${Y1 - 18} Z`, fill: SOFT }, svg);+ [LIL, SKY, MINT].forEach((c, i) => el("circle", { cx: 72 + i * 24, cy: 114, r: 7, fill: c, stroke: INK, "stroke-width": 2 }, svg));+ /* the switch in the title bar */+ el("rect", { x: 560, y: 98, width: 180, height: 32, rx: 16, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ const thumb = el("rect", { x: 650, y: 101, width: 87, height: 26, rx: 13, fill: YEL, stroke: INK, "stroke-width": 2 }, svg);+ txt(svg, 605, 120, "Classic", { "text-anchor": "middle", "font-size": 17, "font-weight": 600, fill: MUTE });+ txt(svg, 693, 120, "PRO", { "text-anchor": "middle", "font-size": 17, "font-weight": 700 });+ /* the pane: a slot for each row's line, dashed while it is empty */+ const notes = NOTE.map((s, i) => {+ const bar = el("rect", { x: NX - 14, y: RY[i] - 21, height: 42, rx: 12, fill: SOFT, stroke: LIL, "stroke-width": 2, "stroke-dasharray": "6 6" }, svg);+ el("circle", { cx: 278, cy: RY[i], r: 5, fill: LIL, stroke: INK, "stroke-width": 2 }, svg);+ const n = fit(txt(svg, NX, RY[i] + 7, s, { "font-size": 21, "font-weight": 600 }), 420), w = n.getComputedTextLength();+ bar.setAttribute("width", (w + 28).toFixed(1));+ const id = `mg-sidebar-note-${i}`, cp = el("clipPath", { id }, svg);+ const cr = el("rect", { x: NX - 4, y: RY[i] - 24, width: 0, height: 48 }, cp);+ n.setAttribute("clip-path", `url(#${id})`);+ return { n, bar, cr, w };+ });+ /* the highlight, then the rows; the Stapler row's shadow shows when it floats */+ const shadow = el("ellipse", { cx: 153, cy: 487, rx: 78, ry: 3.5, fill: LIL }, svg);+ const hl = el("rect", { x: 54, y: -23, width: 198, height: 46, rx: 12, fill: YEL, stroke: INK, "stroke-width": 2 }, svg);+ const ICON = [+ (p, x, y) => { el("rect", { x: x - 11, y: y - 10, width: 22, height: 20, rx: 4, fill: SKY, stroke: INK, "stroke-width": 2 }, p); el("path", { d: `M${x - 4} ${y - 10} V${y + 10} M${x + 4} ${y - 10} V${y + 10}`, stroke: INK, "stroke-width": 2 }, p); },+ (p, x, y) => el("path", { d: `M${x - 10} ${y - 10} h20 a3 3 0 0 1 3 3 v10 a3 3 0 0 1 -3 3 h-11 l-7 6 v-6 h-2 a3 3 0 0 1 -3 -3 v-10 a3 3 0 0 1 3 -3 z`, fill: LIL, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, p),+ (p, x, y) => el("path", { d: `M${x + 3} ${y - 12} L${x - 8} ${y + 2} H${x} L${x - 3} ${y + 12} L${x + 8} ${y - 2} H${x} Z`, fill: YEL, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, p),+ (p, x, y) => { el("path", { d: `M${x - 8} ${y + 6} L${x} ${y - 7} L${x + 8} ${y + 6} Z`, fill: "none", stroke: INK, "stroke-width": 2 }, p); [[-8, 6], [0, -7], [8, 6]].forEach(([a, b]) => el("circle", { cx: x + a, cy: y + b, r: 4.5, fill: MINT, stroke: INK, "stroke-width": 2 }, p)); },+ (p, x, y) => star(p, x, y, 12, BLUE),+ (p, x, y) => { el("rect", { x: x - 13, y: y + 3, width: 26, height: 7, rx: 3.5, fill: "#fff", stroke: INK, "stroke-width": 2 }, p); el("rect", { x: x - 13, y: y - 8, width: 24, height: 8, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2, transform: `rotate(-8 ${x + 10} ${y - 4})` }, p); el("circle", { cx: x + 10, cy: y + 1, r: 3, fill: INK }, p); },+ ];+ const rows = NAMES.map((s, i) => {+ const g = el("g", {}, svg), inner = el("g", {}, g);+ el("rect", { x: 50, y: RY[i] - 26, width: 206, height: 52, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, inner);+ ICON[i](inner, 80, RY[i]);+ fit(txt(inner, 104, RY[i] + 7, s, { "font-size": 21, "font-weight": 700 }), 142);+ button(g, `Show what ${s} is for`, () => api.seek(LIT[i] - 0.5));+ return inner;+ });+ /* the window's edge and lines go on last, so the border stays even */+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "none", stroke: INK, "stroke-width": 2.5, "pointer-events": "none" }, svg);+ el("path", { d: `M${X0} ${YB} H${X1} M${SW} ${YB} V${Y1}`, stroke: INK, "stroke-width": 2, "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the PRO side of the switch gives one small press at the start */+ const pr = 1 - 0.08 * Math.sin(Math.PI * seg(t, 0.2, 0.5));+ thumb.setAttribute("transform", `translate(693.5 114) scale(${pr.toFixed(3)}) translate(-693.5 -114)`);+ /* the Stapler row lifts off and floats once it is picked */+ const fl = io(seg(t, LIT[5], LIT[5] + 0.5)) * (1 - out), fy = -9 * fl + 2.2 * fl * Math.sin(now * 2.4);+ rows[5].setAttribute("transform", `translate(0 ${fy.toFixed(2)}) rotate(${(-1.5 * fl).toFixed(2)} 153 462)`);+ shadow.setAttribute("opacity", (0.85 * fl).toFixed(2));+ shadow.setAttribute("transform", `translate(153 0) scale(${(1 - 0.04 * Math.sin(now * 2.4) * fl).toFixed(3)} 1) translate(-153 0)`);+ /* the highlight walks down the rows, landing with a small overshoot */+ let hy = RY[0], on5 = 0;+ for (let j = 1; j < 6; j++) { const k = seg(t, LIT[j] - 0.45, LIT[j]); if (k > 0) hy = lerp(RY[j - 1], RY[j], land(k, 1.6)); if (j === 5) on5 = k; }+ const hs = Math.min(1.15, land(seg(t, 0.5, 0.85), 2)) * (1 - out);+ hl.setAttribute("transform", `translate(153 ${(hy + fy * on5).toFixed(2)}) rotate(${(-1.5 * fl).toFixed(2)}) scale(${hs.toFixed(3)}) translate(-153 0)`);+ hl.setAttribute("opacity", hs > 0.01 ? 1 : 0);+ /* each row writes its line; the current one keeps its soft slot */+ notes.forEach((nn, j) => {+ const k = oc(seg(t, LIT[j], LIT[j] + 0.6)) * (1 - out), cur = t >= LIT[j] && (j === 5 || t < LIT[j + 1]) && t < 13.3;+ nn.cr.setAttribute("width", ((nn.w + 8) * k).toFixed(1));+ nn.n.setAttribute("opacity", k > 0.01 ? 1 : 0);+ nn.bar.setAttribute("stroke-opacity", k > 0.4 ? 0 : 1);+ nn.bar.setAttribute("fill", cur || k < 0.4 ? SOFT : "#fff");+ });+ say(cap, t < 0.9 ? "The switch in the title bar moves between Classic and PRO" : "One window with a sidebar, and a screen for each part of the office", 700);+ };+ },+ };++ /* ---- askme: Ask me questions wait in a queue; the one in front gets a one line answer, its question mark turns+ into a tick, it moves on and the queue shuffles up (the last card is a little too keen and bumps the one ahead) ---- */+ SC.askme = {+ dur: 14, still: 11.0,+ build(svg, api) {+ head(svg, "Answering unblocks the task", "tap the card");+ const BX = 52, BY = 124, BW = 424, BH = 330, QX = 630, QY = [214, 326, 438], QW = 196, QH = 90, QB = [306, 280], FY = BY + 217;+ txt(svg, QX, 152, "waiting on a person", { "text-anchor": "middle", "font-size": 18, "font-weight": 600, fill: MUTE });+ const qcard = () => {+ const g = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: -QW / 2, y: -QH / 2, width: QW, height: QH, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-46 -14 H72 M-46 12 H40", stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.55 }, g);+ el("circle", { cx: -72, cy: -1, r: 15, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ txt(g, -72, 6, "?", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ return g;+ };+ const Q = [0, 1, 2, 3].map(qcard);+ /* the card that grows from the front of the queue into the big card */+ const mg = el("g", { "pointer-events": "none" }, svg);+ const mr = el("rect", { fill: "#fff", stroke: INK, "stroke-width": 2.5 }, mg);+ const ml = el("path", { stroke: MUTE, "stroke-linecap": "round", "stroke-opacity": 0.55 }, mg);+ const mb = el("g", {}, mg);+ el("circle", { r: 15, fill: LIL, stroke: INK, "stroke-width": 2.5 }, mb);+ txt(mb, 0, 7, "?", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ /* the big Ask me card */+ const BG = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: BX, y: BY, width: BW, height: BH, rx: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, BG);+ const chip = pill(BG, "Ask me", YEL);+ tf(chip, BX + 24 + parseFloat(chip.firstChild.getAttribute("width")) / 2, BY + 46);+ txt(BG, BX + BW - 26, BY + 53, "Inbox", { "text-anchor": "end", "font-size": 17, "font-weight": 600, fill: MUTE });+ el("path", { d: `M${BX + 30} ${BY + 100} H${BX + 372} M${BX + 30} ${BY + 128} H${BX + 342} M${BX + 30} ${BY + 156} H${BX + 220}`, stroke: MUTE, "stroke-width": 9, "stroke-linecap": "round", "stroke-opacity": 0.55 }, BG);+ const qb = el("g", {}, BG);+ el("circle", { r: 18, fill: LIL, stroke: INK, "stroke-width": 2.5 }, qb);+ txt(qb, 0, 8, "?", { "text-anchor": "middle", "font-size": 23, "font-weight": 700 });+ const tb = el("g", {}, BG);+ el("circle", { r: 18, fill: MINT, stroke: INK, "stroke-width": 2.5 }, tb);+ el("path", { d: "M-8 0 l5 6 l11 -12", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, tb);+ const field = el("rect", { x: BX + 24, y: FY - 27, width: BW - 48, height: 54, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 2 }, BG);+ const ph = txt(BG, BX + 44, FY + 6, "Answer in one line", { "font-size": 20, "font-weight": 500, fill: MUTE });+ const typed = el("path", { d: `M${BX + 46} ${FY} H${BX + 110} M${BX + 124} ${FY} H${BX + 186}`, stroke: INK, "stroke-width": 9, "stroke-linecap": "round", pathLength: 1, "stroke-dasharray": "0 1" }, BG);+ const caret = el("path", { d: `M0 ${FY - 15} V${FY + 15}`, stroke: BLUE, "stroke-width": 2.5, "stroke-linecap": "round" }, BG);+ const send = el("g", {}, BG);+ el("circle", { r: 18, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, send);+ el("path", { d: "M-7 0 H6 M1 -6 L7 0 L1 6", fill: "none", stroke: "#fff", "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, send);+ const wait = pill(BG, "waiting on you", SKY);+ const ww = parseFloat(wait.firstChild.getAttribute("width"));+ const unb = pill(BG, "unblocks the task", MINT, true);+ const uw = parseFloat(unb.firstChild.getAttribute("width"));+ const hit = el("g", {}, svg);+ el("rect", { x: BX - 7, y: BY - 7, width: BW + 14, height: BH + 14, rx: 27, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Ask again", () => api.seek(0.3));+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ /* two questions per loop: the first moves on at 3.3, the second holds until 12.6 */+ const L0 = t < 8 ? 3.3 : 12.6, M0 = L0 + 0.5;+ const lk = io(seg(t, L0, L0 + 0.5));+ const bo = t < L0 ? 1 : t < L0 + 0.5 ? 1 - lk : seg(t, L0 + 1.1, L0 + 1.25);+ const bx = t >= L0 && t < L0 + 0.5 ? -70 * lk : 0, br = t >= L0 && t < L0 + 0.5 ? -4 * lk : 0;+ BG.setAttribute("transform", `translate(${bx.toFixed(2)} 0) rotate(${br.toFixed(2)} ${BX + BW / 2} ${BY + BH / 2})`);+ BG.setAttribute("opacity", bo.toFixed(2));+ /* the answer: focus, type one short line, send */+ const ty = t < 3.8 ? seg(t, 0.9, 2.1) : t < 13.1 ? seg(t, 5.9, 7.1) : 0;+ const focus = (t > 0.7 && t < 2.3) || (t > 5.7 && t < 7.3);+ const tick = (t >= 2.5 && t < 3.8) || (t >= 7.5 && t < 13.1), tk = t < 5 ? seg(t, 2.5, 2.75) : seg(t, 7.5, 7.75);+ const press = Math.sin(Math.PI * seg(t, 2.25, 2.45)) + Math.sin(Math.PI * seg(t, 7.25, 7.45));+ field.setAttribute("stroke", focus ? BLUE : INK);+ ph.setAttribute("opacity", ty > 0 ? 0 : 1);+ typed.setAttribute("stroke-dasharray", `${ty.toFixed(3)} 1`);+ typed.setAttribute("opacity", ty > 0.005 ? 1 : 0);+ const len = ty * 126, cx = len <= 64 ? BX + 46 + len : BX + 124 + (len - 64);+ caret.setAttribute("transform", `translate(${(cx + (ty > 0 ? 9 : 0)).toFixed(2)} 0)`);+ caret.setAttribute("opacity", focus && Math.floor(now * 2.6) % 2 === 0 ? 1 : 0);+ tf(send, BX + BW - 50, FY, (1 - 0.14 * press).toFixed(3));+ const breathe = tick ? 0 : 0.05 * Math.sin(now * 3);+ tf(qb, QB[0], QB[1], (tick ? 1 - oc(tk) : 1 + breathe).toFixed(3));+ qb.setAttribute("opacity", tick && tk >= 1 ? 0 : 1);+ tf(tb, QB[0], QB[1], (tick ? Math.min(1.25, land(tk, 2.4)) : 0).toFixed(3));+ tb.setAttribute("opacity", tick ? 1 : 0);+ const wk = tick ? io(seg(tk, 0, 0.5)) : 0, uk = tick ? Math.min(1.2, land(seg(tk, 0.4, 1), 2.2)) : 0;+ tf(wait, BX + 24 + ww / 2, BY + 288, (1 - wk).toFixed(3));+ wait.setAttribute("opacity", wk >= 1 ? 0 : 1);+ tf(unb, BX + 24 + uw / 2, BY + 288, uk.toFixed(3));+ unb.setAttribute("opacity", uk > 0.01 ? 1 : 0);+ /* the big card grows out of the front of the queue */+ const mk = io(seg(t, M0, M0 + 0.65)), mvis = t >= M0 && t < M0 + 0.8;+ const x0 = QX - QW / 2, y0 = QY[0] - QH / 2;+ mr.setAttribute("x", lerp(x0, BX, mk).toFixed(2)); mr.setAttribute("y", lerp(y0, BY, mk).toFixed(2));+ mr.setAttribute("width", lerp(QW, BW, mk).toFixed(2)); mr.setAttribute("height", lerp(QH, BH, mk).toFixed(2));+ mr.setAttribute("rx", lerp(16, 22, mk).toFixed(2));+ const lx = lerp(x0 + 52, BX + 30, mk), l1 = lerp(x0 + 170, BX + 372, mk), l2 = lerp(x0 + 138, BX + 342, mk), ly1 = lerp(y0 + 31, BY + 100, mk), ly2 = lerp(y0 + 57, BY + 128, mk);+ ml.setAttribute("d", `M${lx.toFixed(1)} ${ly1.toFixed(1)} H${l1.toFixed(1)} M${lx.toFixed(1)} ${ly2.toFixed(1)} H${l2.toFixed(1)}`);+ ml.setAttribute("stroke-width", lerp(7, 9, mk).toFixed(2));+ tf(mb, lerp(QX - 72, QB[0], mk).toFixed(2), lerp(QY[0] - 1, QB[1], mk).toFixed(2), lerp(1, 1.2, mk).toFixed(3));+ mb.setAttribute("opacity", t < L0 + 1.1 ? 1 : 0);+ mg.setAttribute("opacity", mvis ? 1 : 0);+ /* the queue shuffles up; on the first shuffle the last card is too keen and bumps the one ahead */+ const pos = [[QX, QY[0], 0, 1], [QX, QY[1], 0, 1], [QX, QY[2], 0, 1], [QX + 80, QY[2], 0, 0]];+ const SH = t >= M0 && t < M0 + 1.4 ? [M0, L0 < 8 ? 1 : 0] : null;+ if (SH) {+ const [s0, bump] = SH, k1 = seg(t, s0, s0 + 0.5), k2 = seg(t, s0 + 0.15, s0 + 0.7), k3 = seg(t, s0 + 0.4, s0 + 0.9), hb = t - (s0 + 0.42);+ const nudge = bump && hb > 0 ? Math.exp(-5 * hb) * Math.sin(16 * hb) : 0;+ pos[0][3] = 0;+ pos[1] = [QX, lerp(QY[1], QY[0], land(k1, 1.3)) - 7 * nudge, 4 * nudge, 1];+ pos[2] = [QX, lerp(QY[2], QY[1], land(k2, bump ? 2.8 : 1.3)), -3 * nudge, 1];+ pos[3] = [lerp(QX + 80, QX, oc(k3)), QY[2], 0, cl(k3 * 3)];+ }+ Q.forEach((g, j) => { tf(g, pos[j][0].toFixed(2), pos[j][1].toFixed(2), 1, pos[j][2].toFixed(2)); g.setAttribute("opacity", pos[j][3].toFixed(2)); });+ say(cap, t < 0.7 ? "Ask me is the list of questions waiting on a person" : t < 2.3 ? "Answer in one line" : t < 3.3 ? "Answering unblocks the task" : t < 5.7 ? "and the card moves on" : t < 7.3 ? "Answer in one line" : t < 12.6 ? "A short answer unblocks the task, and the card moves on" : "This is the screen that keeps an office moving", 700);+ };+ },+ };++ /* ---- stapler: Show the stapler goes on, the round stapler pops up over a plain app, its ring opens, Screenshot takes+ a region, a note is added and stapled to it, and the pair goes to an agent; then the ring opens again ---- */+ SC.stapler = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Staple it", "tap the stapler");+ const SX = 680, SY = 408, PCX = 335, PCY = 282, AX = 88, AY = 108;+ /* the setting */+ el("rect", { x: 448, y: 82, width: 308, height: 48, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ fit(txt(svg, 468, 113, "Show the stapler", { "font-size": 20, "font-weight": 700 }), 200);+ const trk = el("rect", { x: 684, y: 93, width: 52, height: 26, rx: 13, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ const knob = el("circle", { cy: 106, r: 8.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ /* where captures go */+ const ag = el("g", {}, svg);+ const agR = el("rect", { x: 62, y: 88, height: 40, rx: 20, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ag);+ const agT = txt(ag, 112, 115, "an agent", { "font-size": 20, "font-weight": 700 });+ agR.setAttribute("width", (agT.getComputedTextLength() + 72).toFixed(1));+ el("circle", { cx: AX, cy: AY, r: 12, fill: SKY, stroke: INK, "stroke-width": 2.5 }, ag);+ const agK = el("g", {}, ag);+ el("circle", { cx: AX, cy: AY, r: 12, fill: MINT, stroke: INK, "stroke-width": 2.5 }, agK);+ el("path", { d: `M${AX - 6} ${AY} l4 4.5 l8 -9`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, agK);+ /* a plain app window with grey bars */+ el("rect", { x: 44, y: 146, width: 712, height: 346, rx: 16, fill: "#fff" }, svg);+ el("path", { d: "M44 182 H180 V492 H60 Q44 492 44 476 Z", fill: SOFT }, svg);+ [LIL, SKY, MINT].forEach((c, i) => el("circle", { cx: 68 + i * 22, cy: 164, r: 6.5, fill: c, stroke: INK, "stroke-width": 2 }, svg));+ el("rect", { x: 330, y: 158, width: 140, height: 12, rx: 6, fill: SOFT }, svg);+ const bars = (d) => el("path", { d, stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.45 }, svg);+ bars("M66 214 H150 M66 244 H136 M66 274 H154 M66 304 H128 M66 334 H146");+ [[202, 202, 264, 124], [486, 202, 250, 124], [202, 346, 534, 126]].forEach(([x, y, w, h]) => el("rect", { x, y, width: w, height: h, rx: 12, fill: SOFT }, svg));+ bars("M226 236 H400 M226 262 H440 M226 288 H350 M510 236 H690 M510 262 H650 M510 288 H700 M226 380 H600 M226 406 H520 M226 432 H640");+ el("rect", { x: 380, y: 276, width: 64, height: 38, rx: 8, fill: SKY }, svg);+ el("rect", { x: 44, y: 146, width: 712, height: 346, rx: 16, fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M44 182 H756", stroke: INK, "stroke-width": 2 }, svg);+ /* the region box drawn by the crosshair */+ const box = el("rect", { x: 212, y: 210, width: 0, height: 0, rx: 4, fill: SKY, "fill-opacity": 0.18, stroke: BLUE, "stroke-width": 2.5, "stroke-dasharray": "8 6" }, svg);+ /* the capture with its note, stapled; it flies as one */+ const P = mv(el("g", { "pointer-events": "none" }, svg)), Pin = el("g", { transform: `translate(${-PCX} ${-PCY})` }, P);+ el("rect", { x: 212, y: 210, width: 246, height: 108, rx: 10, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, Pin);+ el("path", { d: "M230 236 H400 M230 262 H436 M230 288 H350", stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.45 }, Pin);+ el("rect", { x: 380, y: 270, width: 60, height: 36, rx: 8, fill: SKY }, Pin);+ const note = el("g", {}, Pin);+ el("rect", { x: 212, y: 310, width: 246, height: 46, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, note);+ const nph = txt(note, 232, 340, "your note", { "font-size": 18, "font-weight": 500, fill: MUTE });+ const ntx = el("path", { d: "M234 334 H330 M344 334 H420", stroke: INK, "stroke-width": 7, "stroke-linecap": "round", pathLength: 1, "stroke-dasharray": "0 1" }, note);+ const staple = el("path", { d: "M-12 5 V-4 H12 V5", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, Pin);+ /* the ring: four actions that open out of the stapler */+ const RING = [["Screenshot", 502, 420], ["Record message", 512, 344], ["Record meeting", 566, 274], ["Make invisible", 650, 216]];+ const ICO = [+ (g) => el("path", { d: "M-8 -3 V-8 H-3 M3 -8 H8 V-3 M8 3 V8 H3 M-3 8 H-8 V3", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g),+ (g) => { el("rect", { x: -4, y: -10, width: 8, height: 13, rx: 4, fill: LIL, stroke: INK, "stroke-width": 2 }, g); el("path", { d: "M-7 0 Q-7 7 0 7 Q7 7 7 0 M0 7 V10", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g); },+ (g) => { el("circle", { r: 9.5, fill: "none", stroke: INK, "stroke-width": 2 }, g); el("circle", { r: 5.5, fill: BLUE }, g); },+ (g) => el("circle", { r: 8, fill: "none", stroke: MUTE, "stroke-width": 2.5, "stroke-dasharray": "3.5 3.5" }, g),+ ];+ const pills = RING.map(([s, x, y], i) => {+ const g = el("g", { "pointer-events": "none" }, svg);+ const r = el("rect", { y: -22, height: 44, rx: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ const tx = txt(g, 0, 7, s, { "font-size": 20, "font-weight": 700 }), w = tx.getComputedTextLength() + 70;+ r.setAttribute("x", (-w / 2).toFixed(1)); r.setAttribute("width", w.toFixed(1)); tx.setAttribute("x", (-w / 2 + 50).toFixed(1));+ const ic = el("g", { transform: `translate(${(-w / 2 + 25).toFixed(1)} 0)` }, g);+ el("circle", { r: 16, fill: SOFT }, ic);+ ICO[i](ic);+ return { g, x, y, r };+ });+ /* the stapler: a dashed spot until the setting is on, then a round button that floats */+ const HG = el("g", {}, svg);+ const ghost = el("circle", { cx: SX, cy: SY, r: 34, fill: "#fff", "fill-opacity": 0, stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 7" }, HG);+ el("circle", { cx: SX, cy: SY, r: 42, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, HG);+ button(HG, "Open the stapler ring", () => api.seek(1.45));+ const shadow = el("ellipse", { cx: SX, cy: SY + 50, rx: 26, ry: 5, fill: LIL, "pointer-events": "none" }, svg);+ const B = el("g", { "pointer-events": "none" }, svg);+ el("circle", { r: 34, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, B);+ el("rect", { x: -23, y: 9, width: 46, height: 7, rx: 3.5, fill: SOFT, stroke: INK, "stroke-width": 2 }, B);+ el("rect", { x: 11, y: -5, width: 11, height: 17, rx: 3, fill: INK }, B);+ const arm = el("path", { d: "M18 -9 H-15 Q-23 -9 -23 -2 Q-23 3 -17 3 H18 Z", fill: YEL, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, B);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const xh = el("path", { d: "M-13 0 H13 M0 -13 V13", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ const KP = [[0, 560, 300], [0.15, 560, 300], [0.45, 700, 110], [0.7, 700, 110], [1.35, 686, 412], [1.75, 686, 412], [2.3, 600, 470], [2.65, 444, 432], [3.1, 444, 432]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the setting goes on, and the stapler pops up in its spot */+ const on = io(seg(t, 0.5, 0.75)) * (1 - io(seg(t, 13.3, 13.6)));+ knob.setAttribute("cx", (697 + 26 * on).toFixed(2));+ trk.setAttribute("fill", on > 0.5 ? MINT : SOFT);+ const pop = Math.min(1.2, land(seg(t, 0.75, 1.1), 2.2)) * (1 - io(seg(t, 13.4, 13.7)));+ const press = Math.sin(Math.PI * seg(t, 1.45, 1.65)), bobY = 3 * Math.sin(now * 2) * cl(pop);+ tf(B, SX, (SY + bobY).toFixed(2), (pop * (1 - 0.1 * press)).toFixed(3));+ B.setAttribute("opacity", pop > 0.01 ? 1 : 0);+ ghost.setAttribute("stroke-opacity", (1 - cl(pop)).toFixed(2));+ shadow.setAttribute("rx", (26 * cl(pop) * (1 - 0.06 * Math.sin(now * 2))).toFixed(2));+ shadow.setAttribute("opacity", (0.5 * cl(pop)).toFixed(2));+ /* the staple: the stapler's arm comes down, and a staple joins the note to the capture */+ const chomp = Math.sin(Math.PI * seg(t, 6.85, 7.1));+ arm.setAttribute("transform", `rotate(${(9 - 11 * chomp).toFixed(2)} 16 -3)`);+ /* the ring opens, Screenshot is picked, it closes; later it opens again and stays */+ pills.forEach((p, i) => {+ const o1 = land(seg(t, 1.65 + 0.07 * i, 2.05 + 0.07 * i), 1.6), c1 = i === 0 ? io(seg(t, 3.05, 3.3)) : io(seg(t, 2.9, 3.15));+ const o2 = land(seg(t, 9.3 + 0.08 * i, 9.75 + 0.08 * i), 1.6), c2 = io(seg(t, 13.2, 13.5));+ const k = t < 6 ? o1 * (1 - c1) : o2 * (1 - c2), pick = i === 0 ? Math.sin(Math.PI * seg(t, 2.8, 3.0)) : 0;+ tf(p.g, lerp(SX, p.x, cl(k)).toFixed(2), lerp(SY, p.y, cl(k)).toFixed(2), (Math.min(1.12, Math.max(0, k)) * (1 + 0.08 * pick)).toFixed(3));+ p.g.setAttribute("opacity", k > 0.02 ? 1 : 0);+ p.r.setAttribute("fill", i === 0 && t > 2.8 && t < 3.3 ? YEL : "#fff");+ });+ /* the pointer: the setting, the stapler, Screenshot; then the crosshair drags the region */+ const [px, py] = at(t), click = Math.max(Math.sin(Math.PI * seg(t, 0.5, 0.7)), press, Math.sin(Math.PI * seg(t, 2.8, 3.0)));+ ptr.setAttribute("transform", `translate(${(px + 2 * click).toFixed(2)} ${(py + 3 * click).toFixed(2)})`);+ ptr.setAttribute("opacity", t < 3.1 ? 1 : 0);+ const xk = io(seg(t, 3.1, 3.6)), dk = io(seg(t, 3.7, 4.7));+ const xx = t < 3.7 ? lerp(444, 212, xk) : lerp(212, 458, dk), xy = t < 3.7 ? lerp(432, 210, xk) : lerp(210, 318, dk);+ tf(xh, xx.toFixed(2), xy.toFixed(2));+ xh.setAttribute("opacity", t >= 3.1 && t < 4.9 ? 1 : 0);+ box.setAttribute("width", (246 * dk).toFixed(2)); box.setAttribute("height", (108 * dk).toFixed(2));+ box.setAttribute("opacity", t >= 3.7 && t < 5.0 ? (1 - seg(t, 4.85, 5.0)).toFixed(2) : 0);+ /* the capture lifts, the note slides in and is typed, the staple goes in, the pair flies to the agent */+ const lift = land(seg(t, 4.85, 5.15), 2), nk = oc(seg(t, 5.05, 5.45)), ty = seg(t, 5.7, 6.7);+ note.setAttribute("transform", `translate(0 ${(-30 * (1 - nk)).toFixed(2)})`);+ note.setAttribute("opacity", nk.toFixed(2));+ nph.setAttribute("opacity", ty > 0 ? 0 : 1);+ ntx.setAttribute("stroke-dasharray", `${ty.toFixed(3)} 1`);+ ntx.setAttribute("opacity", ty > 0.005 ? 1 : 0);+ const sk = seg(t, 6.95, 7.15);+ tf(staple, 234, 312, t < 6.95 ? 0 : Math.min(1.5, land(sk, 2.4)).toFixed(3));+ const fk = io(seg(t, 7.5, 8.4)), jig = t > 6.95 && t < 7.4 ? 2.5 * Math.sin((t - 6.95) * 40) * (1 - seg(t, 6.95, 7.4)) : 0;+ const fx = lerp(PCX, AX, fk), fy = lerp(PCY, AY, fk) + 24 * Math.sin(Math.PI * fk);+ tf(P, fx.toFixed(2), fy.toFixed(2), ((1 + 0.03 * (lift - cl(lift))) * (1 - 0.9 * fk)).toFixed(3), (jig - 8 * Math.sin(Math.PI * fk)).toFixed(2));+ const po = t < 4.85 || t > 8.4 ? 0 : 1 - seg(fk, 0.6, 0.85);+ P.setAttribute("opacity", po.toFixed(2));+ P.setAttribute("display", po < 0.01 ? "none" : "inline");+ /* the agent takes it, gives a small hop and a tick */+ const ak = seg(t, 8.35, 8.6), hop = 6 * Math.sin(Math.PI * seg(t, 8.35, 8.65));+ ag.setAttribute("transform", `translate(0 ${(-hop).toFixed(2)})`);+ tf(agK, 0, 0, 1);+ agK.setAttribute("opacity", t < 8.35 ? 0 : Math.min(1, ak * 3 * (1 - out)).toFixed(2));+ say(cap, t < 0.7 || t >= 13.3 ? "Turn on Show the stapler" : t < 1.55 ? "The Stapler is a small window that floats above every app" : t < 2.9 ? "Then click it for its ring" : t < 5.0 ? "Screenshot captures a region of the screen" : t < 6.9 ? "and sends it to an agent with a note" : t < 9.2 ? "When you see a bug in any app, do not describe it. Staple it" : "Its ring offers Screenshot, Record message, Record meeting and Make invisible", 700);+ };+ },+ };++ /* ==== ADD NEW SCENES ABOVE THIS LINE ==== */++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.pngnew file mode 100644index 00000000..0107114eBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/schedule.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/schedule.pngnew file mode 100644index 00000000..434b5b08Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/schedule.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/sidebar.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/sidebar.pngnew file mode 100644index 00000000..ff83c9ffBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/sidebar.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/slack.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/slack.pngnew file mode 100644index 00000000..1f8ed85cBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/slack.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/stapler.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/stapler.pngnew file mode 100644index 00000000..017c71f0Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/stapler.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/team.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/team.pngnew file mode 100644index 00000000..ce41aa9dBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/team.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/temp.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/temp.pngnew file mode 100644index 00000000..f91cd899Binary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/temp.png differdiff --git a/blog/src/assets/media/your-first-hour-with-munder-difflin/tokens.png b/blog/src/assets/media/your-first-hour-with-munder-difflin/tokens.pngnew file mode 100644index 00000000..4cd452bfBinary files /dev/null and b/blog/src/assets/media/your-first-hour-with-munder-difflin/tokens.png differdiff --git a/blog/src/posts/ai-agent-use-cases-our-office.md b/blog/src/posts/ai-agent-use-cases-our-office.mdnew file mode 100644index 00000000..194a4761--- /dev/null+++ b/blog/src/posts/ai-agent-use-cases-our-office.md@@ -0,0 +1,150 @@+---+title: "AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does"+seoTitle: "AI Agent Use Cases: The Jobs Our Own Agents Do Every Day"+description: "We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it."+date: 2026-10-14+category: guides+categoryLabel: Guides+type: Non-technical+pinned: true+pinOrder: 3+startHere: true+startOrder: 3+primaryKeyword: "ai agent use cases"+secondaryKeywords: ["multi agent use cases", "ai agent team examples", "munder difflin use cases", "ai agents for a small company", "what can ai agents do", "ai agent roles", "ai agent workflows"]+tags: ["Guides", "Multi-Agent", "Getting Started", "AI Agents"]+faq:+ - q: "How many AI agents do you need to start?"+ a: "One. Every office starts with an orchestrator, the agent you talk to. Add a second agent when a job repeats, and a third when you want one agent to check another's work. Ours has fourteen today."+ - q: "Do all the agents run on the same model?"+ a: "No. Each agent picks its own engine and model when you add it. On our floor the research scout runs on Antigravity and the others run on Claude Code."+ - q: "Can the agents send email or merge code on their own?"+ a: "Email, no: it is written as a draft and never sent. Code, partly: our engineering lead merges into a release branch but never into main, and pull requests to this website wait for a person. Community triage adds labels and closes nothing. We set those limits in each agent's brief."+ - q: "What is a temp?"+ a: "A temp is a short lived agent started for one job. It gets an objective and a working folder, does the job, reports done and is removed. A lasting agent keeps its memory and its inbox; a temp's are filed away when it ends."+ - q: "Does the office keep working when the laptop is closed?"+ a: "No. The agents, schedules and connections run on your own computer, so they stop when it sleeps. Two of our scheduled agents check for mains power first and skip the run on battery."+---++We run our own company on [Munder Difflin](https://harnessmd.com/download), and these are the AI agent use cases on our floor: fourteen agents, each with one job, one way its work starts and one thing it hands back. Checked against our own records on 14 Oct 2026.++Munder Difflin is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer. New here? The [install guide](/blog/how-to-install-and-use-munder-difflin/) gets you a working office, [your first hour](/blog/your-first-hour-with-munder-difflin/) covers the first jobs, and the [Guides hub](/blog/topics/guides/) has the rest.++## What does the whole floor look like?++Fourteen agents, and each seat has one job. Most names come from the cast the app ships with; BOB, AgentG and Darryl are names we chose.++| Seat | Agent | The job in one line | How its work starts |+| --- | --- | --- | --- |+| Orchestrator | your clone | Takes your orders, splits them into jobs, asks you only what needs you | Your chat |+| Mission control | Ryan | Keeps one page of company numbers fresh | A schedule, every 6 hours |+| Community lead | Angela | Sorts what users say into a digest and a demand list | When the orchestrator asks |+| Growth | Jim | Researches and writes this blog | Schedules at 09:30 and 21:30 |+| Motion graphics | BOB | Launch videos and animation loops from real app screens | A request |+| Research | AgentG | A daily scout report, a link for every claim | One message from Jim |+| Founder's office | Andy | A morning briefing, Slack replies, first drafts | A schedule at 09:00, and Slack |+| Scrum master | Meredith | Keeps the task board readable | A schedule, every 12 hours |+| Engineering lead | Kevin | Plans the work and gates every piece | A builder saying "done" |+| Engineers | Darryl, Stanley, Creed, Dwight | Build the pieces | Kevin's orders (Dwight: the founder's) |+| Designer | Pam | Draws every new screen before any code | A task card |++Ad hoc agents and temps fill the gaps, and they get their own section below. Two seats, community and motion graphics, have been quiet since early October.++<figure class="mg" data-scene="floor"><img src="/blog/assets/media/ai-agent-use-cases-our-office/floor.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A floor plan of a small office on white, drawn from above, with eleven desks. Name cards drop onto the desks one by one: Orchestrator, Mission control, Community lead, Growth, Motion graphics, Research, Founder's office, Scrum master, Engineering lead, Engineers, Designer. A small pot plant on each desk grows a leaf as its card lands. The Engineers desk is wide, with four chairs, and one chair at the Research desk is left spinning."><figcaption>Our floor on 14 Oct 2026: eleven kinds of seat, fourteen agents, one job each. The drawing shows the seats as desks seen from above. Tap a desk to read its job and grow its plant a leaf.</figcaption></figure>++And here is part of the real thing, at six in the morning:++{% img "busy-office", "The middle of our real office floor on 14 Oct 2026, in the Classic view of version 0.5.5." %}++## What does the orchestrator do?++The orchestrator is the only agent you have to talk to. Ours takes an order from the founder's chat, writes a short dispatch for the right agent (the objective, the output, the tools, the limits) and reports back when it is done.++It also guards your attention. When only a person can move a job forward, it writes one short question on a card. We call that an ASK ME card. The answer goes back on the same card, and nobody sits idle waiting for it.++**Copy it:** you already have this seat. It is the agent you name during setup. [How the orchestrator routes work](/blog/how-the-god-orchestrator-works/) explains its decisions.++## Which agents work on a clock?++Four of ours do their main job on a schedule, with no person typing anything.++* **Andy, 09:00.** Reads the company mailbox and calendar and writes one briefing as a draft. It never sends it.+* **Jim, 09:30 and 21:30.** Checks search data, reads the scout's report, writes pages and opens one pull request per run. Three shorter checks a day look for new launches.+* **Ryan, every 6 hours.** Wakes on a schedule and, twice a day, runs the scripts that refresh the company dashboard. Asked for a number, it marks each one measured or inferred.+* **Meredith, every 12 hours.** Rewrites each workstream card in one shape: where it stands, who has it, what changed, what is next, what waits on you.++A board nobody can read is a diary. Meredith's has one reader, and that is enough.++<figure class="mg" data-scene="clock"><img src="/blog/assets/media/ai-agent-use-cases-our-office/clock.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A large wall clock with a pendulum on white. As the hand sweeps through one day, four cards slide out from behind the clock, one to each corner: 09:00 briefing, 09:30 and 21:30 blog runs, every 6 hours dashboard, every 12 hours task board. A dot on each card fills each time its agent starts."><figcaption>Four of our agents do their main job on a schedule: a briefing at 09:00, blog runs at 09:30 and 21:30, the dashboard agent every 6 hours and a task board update every 12 hours. Tap the clock to run the day again.</figcaption></figure>++**Copy it:** in the app, schedules live under Automations (Triggers in the classic office). Pick the agent, pick the time, and write the message. It is sent word for word on every run.++## Who writes the code, and who checks it?++Builders build, and a different agent decides whether it is good. That split is the most useful thing on our floor.++1. **Pam draws first.** Every new screen is an HTML page with its loading, empty and error states before anyone writes app code.+2. **Builders build to an order.** Darryl, Stanley and Creed each get a piece from Kevin, with the files and the limits named. Dwight builds the next release directly with the founder.+3. **Kevin is the gate.** He reads the diff, writes his own tests, runs the full suite and merges into a release branch, never into main.+4. **Creed reads second.** Code that guards keys or commands gets a second, independent read.+5. **Pam compares.** The built screen is checked against her drawing, picture by picture.++On our floor the gate writes the tests, not the builder. Nobody grades their own homework.++<figure class="mg" data-scene="gate"><img src="/blog/assets/media/ai-agent-use-cases-our-office/gate.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A drawing board, a workbench and a small drawbridge stand in a row on white. A paper screen design leaves the drawing board, labelled draws first, becomes a jigsaw piece at the workbench, labelled build to an order, and hops up to the raised bridge under a sign reading the gate. Three lights come on one by one: reads the diff, own tests, full suite. The bridge lowers and the piece hops across and clicks into a row of pieces labelled release branch."><figcaption>How a change gets released on our floor: the designer draws it, a builder builds it, and a different agent reads the diff, writes its own tests and runs the full suite before it joins a release branch. Tap the bridge to send another piece.</figcaption></figure>++**Copy it:** add a second agent whose only job is review. [How agents verify their own work](/blog/how-ai-agents-verify-their-own-work/) has the method.++## Who deals with the outside world?++Four agents read what is outside the office, and none of them takes the final step alone.++* **Angela** runs the community queues: GitHub issues, pull requests and discussions, Discord and forums. She writes a digest and a demand list ranked by how many different people asked. On GitHub she can add labels. She cannot close or merge.+* **Andy** answers Slack messages sent to the office, and watches the support mailbox. Replies to email are drafts.+* **AgentG** reads the open web for the daily scout report: signed out, read only, and with a rule to leave a source out when it cannot open it.+* **Jim** opens pull requests to this website. A person merges them.++The pattern: agents read widely and write drafts, and the step you cannot take back stays with a person.++<figure class="mg" data-scene="drafts"><img src="/blog/assets/media/ai-agent-use-cases-our-office/drafts.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three clipboards hang in a row on white, each holding a sheet: email draft, pull request and triage label. The first two sheets have a dotted line to sign. A fountain pen lying on a rest labelled a person lifts, signs the email draft, which gets a tick reading sent, then signs the pull request, which gets a tick reading merged, and lies down again. The third sheet has no line to sign. It collects coloured labels and a tag reading labels only, and one label too many hangs off its edge."><figcaption>What our agents may do alone and what waits for a person: email is drafted and never sent, pull requests to this website wait for a merge, and triage adds labels and closes nothing. The drawing shows a person's approval as a signature. Tap the pen to sign again.</figcaption></figure>++## Which jobs get another engine, or no lasting agent at all?++Not every job deserves a desk.++**A scout on another engine.** AgentG runs on Antigravity while the others run on Claude Code. A scout reads a great deal and decides very little, so it can run on a different engine from the builders. [Model routing](/blog/do-more-with-less-model-routing/) covers the idea.++**Ad hoc agents.** We added a second scout on Grok on 2 Oct 2026 and removed it on 13 Oct 2026, when its usage limits kept running out. It was hired and gone in eleven days, with no farewell card.++**Temps.** A temp is started for one job and removed when it reports done. Jobs ours have done: run the test suite and write an honest report, fix one release bug as a relay of four, each picking up the last one's work, and free disk space on a laptop. That last job took three temps. The first freed nothing, the second ran out of budget, and the third freed 39 GiB.++**Sub agents** are different again: helpers inside one agent's own session. The fact reviewers and scene artists for this blog are sub agents of Jim. They never appear on the floor.++<figure class="mg" data-scene="temps"><img src="/blog/assets/media/ai-agent-use-cases-our-office/temps.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three job tickets stand on poles in a row on white: run the test suite, fix one release bug, free disk space. Beside them is a small house with a smoking chimney and a sign reading lasting agent. A tent arrives as a rolled bag and is pitched under the first ticket, a bar fills, the ticket is marked done, and the tent is packed and slides away. The same happens for the second and third jobs, and the third tent needs three goes to stand up. The house stays where it is."><figcaption>A temp is started for one job and removed when it reports done; a lasting agent keeps its desk, its memory and its inbox. The drawing shows temps as tents and the lasting agent as a house. Tap a job to pitch its tent again.</figcaption></figure>++## How does work move between them?++Through files, on purpose. Every agent has its own folder with an inbox, an outbox and a memory file, and no agent writes in another agent's folder.++* **Messages.** An agent drops a small file in its outbox and the app delivers it. Two kinds of message, inform and done, end a thread, so two agents cannot reply to each other forever.+* **The plan.** One written plan that only the orchestrator edits, and one task board of cards: todo, doing, blocked, done.+* **The circuit breaker.** It watches every agent for loops, error storms and overspending. It steers first, then limits the agent to reading, then stops it.+* **Secrets.** Keys never reach an agent. A local broker makes the outside call for it.+* **Memory.** Each agent's notes are searchable by the others, so a lesson learned once is not learned twice.++<figure class="mg" data-scene="relay"><img src="/blog/assets/media/ai-agent-use-cases-our-office/relay.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two cabinets shaped like folders stand on white, each with two tube ends labelled outbox and inbox. Clear pneumatic tubes join them and cross inside a box in the middle labelled the app. A small file drops into the first cabinet's outbox and becomes a capsule, which shoots through the tube, passes through the app and lands in the second cabinet's inbox. A capsule labelled done travels back the other way to the first inbox. A tick lights on the app and a flap closes over each outbox to show the thread is finished."><figcaption>How one agent asks another for something: a file in its outbox, delivered by the app to the other's inbox. A message marked done or inform ends the thread, so two agents cannot loop. Tap the outbox to send a note.</figcaption></figure>++## How do you copy this office?++Start with three seats, not fourteen: the orchestrator, one builder and one scheduled job. Add a reviewer when the builder's work starts to matter, a scout when you keep looking the same things up, and a board keeper when the cards pile up.++| Seat | A first message to give it |+| --- | --- |+| Briefing | "Every weekday at 09:00, read the open task cards and your inbox. Write me one short briefing: what moved, what is stuck, what needs me. Never send anything outside the office." |+| Reviewer | "When a builder says a piece is done, read the diff, write your own tests and report pass or fail. Never merge to main." |+| Scout | "Each evening, list the five biggest launches in my field from the last 24 hours, with a link for each. If you cannot open a source, leave it out." |+| Board keeper | "Every 12 hours, rewrite each open card as: where it stands, who has it, what changed, what is next, what waits on me." |++Give each agent one job and one clear limit. Ours got useful the day we stopped asking any of them to do everything.++<link rel="stylesheet" href="/blog/assets/media/ai-agent-use-cases-our-office/motion.css"><script defer src="/blog/assets/media/ai-agent-use-cases-our-office/motion.js"></script>diff --git a/blog/src/posts/claude-code-max-plan-tips.md b/blog/src/posts/claude-code-max-plan-tips.mdindex b588d8cd..dde01467 100644--- a/blog/src/posts/claude-code-max-plan-tips.md+++ b/blog/src/posts/claude-code-max-plan-tips.md@@ -6,7 +6,7 @@ category: guidescategoryLabel: Guidestype: Technicalpinned: true-pinOrder: 3+pinOrder: 4primaryKeyword: "how to get the most out of claude max"secondaryKeywords: ["how to make the most of claude max", "claude code max plan tips", "how to maximize claude code usage", "how to reduce claude code token usage", "claude code usage limit reset", "does claude code max have weekly limits"]tags: ["Guides", "Claude Code", "Cost", "Workflow"]diff --git a/blog/src/posts/codex-max-plan-tips.md b/blog/src/posts/codex-max-plan-tips.mdindex 8233230d..edcdd70c 100644--- a/blog/src/posts/codex-max-plan-tips.md+++ b/blog/src/posts/codex-max-plan-tips.md@@ -6,7 +6,7 @@ category: guidescategoryLabel: Guidestype: Technicalpinned: true-pinOrder: 4+pinOrder: 5primaryKeyword: "codex max plan"secondaryKeywords: ["how to make the most of codex", "codex max plan usage", "chatgpt pro codex usage limits", "does codex have weekly limits", "how to check codex limit usage", "how to reduce codex token usage"]tags: ["Guides", "Codex", "Cost", "Model Routing"]diff --git a/blog/src/posts/grok-bot-alternatives.md b/blog/src/posts/grok-bot-alternatives.mdindex 495209ad..1b1ce8d0 100644--- a/blog/src/posts/grok-bot-alternatives.md+++ b/blog/src/posts/grok-bot-alternatives.md@@ -7,7 +7,7 @@ category: comparisonscategoryLabel: Comparisonstype: Non-technicalpinned: true-pinOrder: 5+pinOrder: 6primaryKeyword: "grok bot alternatives"secondaryKeywords: ["grok bot alternative", "grok bot competitors", "gemini spark vs grok bot", "grok bot open source alternative", "grok bot vs chatgpt dots", "grok bot vs claude cowork", "is grok bot free"]tags: ["Comparisons", "Local-First", "Security", "Open Source", "Memory"]diff --git a/blog/src/posts/herdr-alternatives.md b/blog/src/posts/herdr-alternatives.mdindex fdc987ef..5be28afe 100644--- a/blog/src/posts/herdr-alternatives.md+++ b/blog/src/posts/herdr-alternatives.md@@ -6,7 +6,7 @@ category: comparisonscategoryLabel: Comparisonstype: Technicalpinned: true-pinOrder: 6+pinOrder: 7primaryKeyword: "herdr alternatives"secondaryKeywords: ["herdr tmux alternative", "herdr alternatives for windows", "herdr vs tmux", "herdr vs zellij", "is herdr open source"]tags: ["Comparisons", "Multi-Agent", "Claude Code", "Tools", "Open Source"]diff --git a/blog/src/posts/how-to-install-and-use-munder-difflin.md b/blog/src/posts/how-to-install-and-use-munder-difflin.mdindex ddf081ee..6615465d 100644--- a/blog/src/posts/how-to-install-and-use-munder-difflin.md+++ b/blog/src/posts/how-to-install-and-use-munder-difflin.md@@ -1,498 +1,282 @@----title: "How to Install and Use Munder Difflin: A Beginner's Guide"-description: "A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux."+title: "How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office"+seoTitle: "How to Install Munder Difflin 0.5.6: Beginner's Guide"+description: "Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure."date: 2026-06-05-updated: 2026-09-10+updated: 2026-10-14category: guidescategoryLabel: Guidestype: Non-technicalpinned: truepinOrder: 1primaryKeyword: "how to install munder difflin"-secondaryKeywords: ["munder difflin download", "munder difflin setup", "what is a coding agent", "munder difflin windows", "free ai coding agent", "antigravity free", "munder difflin beginner guide"]-tags: ["Guides", "Getting Started", "Tutorial", "Non-Technical", "Automation"]+secondaryKeywords: ["munder difflin download", "munder difflin setup", "install munder difflin windows", "install munder difflin mac", "what is a coding agent", "munder difflin beginner guide", "munder difflin not installed fix"]+tags: ["Guides", "Getting Started", "Tutorial", "Non-Technical"]author:name: Chaitanya Giriinitials: CGfaq:- - q: "Do I need to know how to code to use Munder Difflin?"- a: "No. The onboarding asks whether you are technical or not on its very first screen, and the non-technical path replaces every piece of jargon with plain language. You will open a terminal exactly once, to install and log into your AI engine, and you can copy and paste those commands."- - q: "Does Munder Difflin cost anything?"- a: "The classic app is free and open source, with unlimited local agents. Pro and Teams are optional paid plans. What you may pay for is the AI engine behind it. Antigravity is available at no charge and OpenCode has a free path, so you can run the whole thing without paying anything. If you already pay for ChatGPT or Claude, you can use those subscriptions instead."- - q: "Can these agents really change files on my computer?"- a: "Yes, and that is the point, so it is worth understanding before you start. An agent can read files, write files and run commands in the folders you give it. You choose how much freedom it has during setup, and you can set it to ask permission before every change."+ - q: "Do I need to know how to code to install Munder Difflin?"+ a: "No. You type two or three commands, to install an AI engine and sign in to it, and you can copy and paste them. If you skip that, the app offers to install a missing engine for you the first time an agent starts."+ - q: "Do I have to install an AI engine before the app?"+ a: "No. Setup marks each engine Installed, Installs on first start or Not installed. An engine marked Installs on first start is installed in the agent's own terminal when you press Install, and the app installs Node first if that engine needs it. You still have to sign in to the engine yourself."+ - q: "Do I need an account?"+ a: "Yes. The first launch asks you to sign in through your browser, on the free version as well. The browser does not open by itself: press Open browser to sign in and sign in there. The screen usually moves on by itself; if it does not, paste the code the browser shows you into the app."- q: "Which operating systems does Munder Difflin run on?"- a: "macOS, Windows and Linux. macOS ships as one universal .dmg for Apple Silicon and Intel, Windows 10 and 11 as a setup installer or a portable build, and Linux as an AppImage."- - q: "Is my code or data sent anywhere?"- a: "The app runs on your machine and stores its files there. Your prompts and the files an agent reads do go to whichever AI engine you picked, the same as if you used that tool directly. Anonymous usage telemetry is opt-out and every event is listed publicly in TELEMETRY.md."+ a: "macOS, Windows and Linux. macOS is one universal build for Apple Silicon and Intel, signed and notarized by Apple. Windows 10 and 11, 64 bit, get a setup installer and a portable build. Linux gets a 64 bit AppImage."+ - q: "Can these agents really change files on my computer?"+ a: "Yes, and that is the point, so understand it before you start. An agent starts in the folder you give it and can read files, write files and run commands. It is not locked in there: it can reach anything your account can. Start with a new empty folder and watch what the agents do before you point one at anything you care about."+ - q: "Is the free version a separate download from Pro?"+ a: "No. There is one installer. The free version is the Classic view, and a licence unlocks the Pro view in the same app. On first launch, press Continue with the free version to reach Classic."startHere: truestartOrder: 1----<div class="callout tldr"><span class="ic">TL;DR</span><p>Pick an AI engine based on what you-already pay for (or pick a free one). Install it with one command in the terminal, which is the-only time you will open a terminal. Download Munder Difflin, answer six onboarding screens, and-you have an office of AI workers you can watch on a screen.</p></div>--If you have never used an AI coding tool, most guides start three steps ahead of you. This one does-not. It starts with what these things actually are, then walks the whole way to a working setup.--You will need about twenty minutes.--## First, what is Munder Difflin?--Munder Difflin is a desktop app that runs a small office of AI workers on your computer.--You see an actual office floor on screen, with characters at desks. Each character is an AI agent-working on something you asked for. One of them is your clone, the boss of the floor, who takes what-you ask for, breaks it into jobs, and hands those jobs to the others. You talk to your clone, and-your clone manages everyone else.--The app is free and open source. It runs on your machine, and it keeps its files on your machine. There are-optional paid plans, Pro and Teams, but nothing in this guide needs them.--The thing it is not: a chatbot. You are not sitting there typing and waiting for replies. You give-your clone a job, close the laptop, and come back to work that was done while you were gone.--{% img "note-1" %}--## What is a coding agent, and how can it touch my computer?+<div class="callout tldr"><span class="ic">TL;DR</span><p>Pick one AI engine, by what you already pay for. Download Munder Difflin 0.5.6 and open it. Press <strong>Open browser to sign in</strong> and sign in, name your workspace and your orchestrator, then press <strong>Continue with the free version</strong>. If the engine is missing, the app offers to install it. About twenty minutes.</p></div>-This is the part worth reading slowly, because it is the part that surprises people.+This is how to install Munder Difflin 0.5.6 when you have never used an AI coding tool. Most guides start three steps ahead of you. This one starts with what the thing is, and ends with a working office and a fix for every way setup can fail.-A **coding agent** is an AI that does not just talk back. It can act. Give a normal chatbot a-question and it returns text. Give a coding agent a job and it will read your files, write new ones,-run commands, install things, and check its own work, in a loop, until the job is done or it gets-stuck.+## What is Munder Difflin?-It does this through a program on your computer called a **CLI agent**. CLI stands for command line-interface, which is the text-only way of controlling a computer that predates windows and buttons.-Claude Code, Codex, Antigravity and the others in this guide are all CLI agents. They are the engine.-Munder Difflin is the office that runs a team of them and keeps track of everything around them:-memory, tasks, schedules, files, and when to come ask you something.+[Munder Difflin](https://harnessmd.com/download) is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer.-**So how does it reach your computer?** When you install one of these engines, it runs as a program-with your permissions. Anything you can do from your own account, it can technically do: open your-files, edit them, delete them, run software, reach the internet. It is not sandboxed away from your-machine by default. That is exactly why it is useful, and exactly why you should be deliberate.+In plain words, it is a small office of AI workers. One of them is your orchestrator, the one you talk to. He is called Michael until you rename him. You give Michael a job, he splits it into pieces, hands the pieces to other agents and comes back to you only when something needs your decision.-Three things keep this sane, and Munder Difflin sets all three up during onboarding:+It is not a chatbot. You do not sit there typing and waiting for replies. You brief Michael, go and do something else, and come back to finished work.-1. **You choose which folders it can work in.** Agents work inside the projects you add. A project- is just a folder. Do not point your first one at your entire home directory.-2. **You choose how much freedom it has.** During setup you pick whether agents work on their own or- pause and ask you before changing files and running commands. If you are new, pick ask.-3. **You can watch it happen.** Every agent has a terminal you can open and read, live. Nothing- happens off-screen.+<figure class="mg" data-scene="office"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/office.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. An office seen from the front on white. One wide desk with a screen stands at the front, marked your orchestrator. Behind it, on a step marked other agents, are four smaller desks, each with its own screen. A slip marked your job lands on the front screen, is torn into three smaller slips, and each flies to a different desk at the back, where a progress bar starts to fill. Two of those desks finish with a tick. The third stops part way and sends back a slip marked needs your decision, which lands on the front screen. The fourth desk gets no slip, and its screen runs a bouncing screensaver. A filing cabinet and a water cooler stand at the sides."><figcaption>What Munder Difflin does: you brief one orchestrator, he hands the pieces to other agents, and only the decisions come back to you. Tap the front desk to send another job.</figcaption></figure>-<div class="callout note"><span class="ic">Start safe</span><p>For your first project, make a brand-new empty folder and use that. Watch what the agents do for an hour before you point one at-anything you care about.</p></div>+## What is a coding agent, and what can it touch on your computer?-## Step 1: Which AI engine will you use?+A coding agent is an AI that acts instead of only answering. Ask a chatbot a question and you get text. Give a coding agent a job and it reads your files, writes new ones, runs commands and checks its own work, in a loop, until the job is done or it gets stuck.-Munder Difflin does not include an AI. It drives one you install. Twelve engines are supported, but-you only need one, and the right one is usually decided by what you already pay for.+Each agent runs through a program called an engine: Claude Code, Codex, Antigravity and the others in this guide. The engine does the thinking. Munder Difflin is the office around a team of them: tasks, memory, schedules, messages, and when to come and ask you.-Find yourself below.+An engine runs with your own permissions. Anything you can do from your account, it can do: open files, change them, delete them, run software. That is why it is useful, and why you should be deliberate. Three things keep it sane:-### You do not want to pay anything+1. **You choose the folder.** An agent starts in the folder you give it. It is not locked in there, so do not make your first one your whole home folder.+2. **You choose how much freedom they have.** Whether agents act on their own or stop and ask is a setting: **Settings**, then **Autonomy & Budgets**. A new install starts on autonomous: agents act without asking. Look at it before your first real job.+3. **You can watch.** Every agent has a live terminal you can open and watch.-**Use Antigravity.** It is Google's agentic development platform and it is-[available at no charge](https://antigravity.google/). It ships with a command line tool called-`agy`, which is what Munder Difflin drives. This is the best starting point for most people reading-this guide.+{% img "autonomy-056", "Settings, then Autonomy & Budgets: how much agents do alone, who may add agents, and the circuit breaker. From a preview build of 0.5.6 with sample data." %}-**OpenCode** is the other free path. It is open source, and it lets you bring your own model,-including free and local ones.+<figure class="mg" data-scene="rooms"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/rooms.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. The floor plan of a house on white. Four pale rooms are each labelled as a folder: Documents, Photos, Projects and Downloads. Beside them is a hatched plot with a plus sign, labelled new empty folder. Its walls are drawn in and it becomes a fifth room, the only one drawn in colour. A small desk with a terminal on it drops into that room, a tag reading agent works here appears under it, and three new files appear beside it. The other rooms stay pale, and loose pages keep piling up in Downloads. Tapping a pale room colours it and moves a second desk in."><figcaption>An agent starts in the folder you give it. It is not locked in there: it can reach anything your account can. For your first job, make a new empty folder and use that. Tap a room to give the agent another folder.</figcaption></figure>-<div class="callout note"><span class="ic">If you are in India, on Jio</span><p>Jio is running an-offer that gives eligible users <a href="https://www.jio.com/google-gemini-offer/">Google AI Pro-free for 18 months</a>. The terms on Jio's page: you must be over 18 and on an active unlimited 5G-plan of ₹349 or above, prepaid or postpaid, kept active for the whole period, claimable once per-number. Google AI Pro raises the daily request limits on <strong>Gemini CLI</strong>, which is a-separate supported engine in Munder Difflin. Note that Jio's page does not mention Antigravity, so-treat this as a boost to the Gemini CLI path rather than to Antigravity. Antigravity is free for-everyone anyway. Check Jio's page for current terms, offers change.</p></div>+## What do you need before you install?-### You already pay for ChatGPT+Less than you would think. The app itself needs nothing else installed.-**Use Codex.** Your ChatGPT Plus or Pro subscription covers it. Nothing more to buy.+* **A computer.** macOS on Apple Silicon or Intel, Windows 10 or 11 (64 bit), or a modern 64 bit Linux.+* **An email address.** The first launch asks you to sign in, on the free version as well.+* **One AI engine, signed in.** You can install it now with one command, or let the app install it later. Either way you sign in to the engine yourself, with your own subscription or key.-### You already pay for Claude+Good to have, and you can add them after setup: git, and two small tools called `uv` and MemPalace that let agents search their memory by meaning. The app checks for these, and for Node, under **Settings**, then **Prerequisites**.-**Use Claude Code.** Your Claude Pro or Max subscription covers it. This is the engine the app was-built against first, and it is the most thoroughly tested of the twelve.+{% img "prerequisites-056", "Settings, then Prerequisites: each helper tool shows Found or Not found, and a missing one shows the command that installs it. From a preview build of 0.5.6 with sample data." %}-### The other engines+## Which AI engine should you pick?-Any of these work. Pick one only if you already use it.+Munder Difflin does not include an AI. It drives one you install. It supports twelve engines plus a custom one, and you need exactly one to start. What you already pay for usually decides it.-| Engine | Command | Good reason to pick it |-|---|---|---|-| Gemini CLI | `gemini` | You have Google AI Pro, including via the Jio offer |-| Grok | `grok` | You have an xAI subscription |-| Kimi Code | `kimi` | You use Kimi already |-| Qwen | `qwen` | You want to run a model locally on your own hardware |-| Crush | `crush` | You like Charm's tools |-| Copilot | `copilot` | You have GitHub Copilot |-| Cursor | `cursor-agent` | You have a Cursor subscription |-| pi | `pi` | You use pi already |+* **You pay for Claude:** use **Claude Code**. It is the app's default engine.+* **You pay for ChatGPT:** use **Codex**.+* **You pay for nothing:** use **Antigravity**, Google's coding tool, whose command is `agy`.+* **You already have a key from an AI provider:** use **OpenCode**, which is open source and runs on the provider and model you choose.-You can add more engines later and run several at once, with different agents on different engines.+<figure class="mg" data-scene="engine"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/engine.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three cards lie on a white table, reading You pay for Claude, You pay for ChatGPT and You pay for nothing. Each card turns over in turn to show the engine to pick: use Claude Code with a Claude subscription, use Codex with a ChatGPT subscription, and use Antigravity if you pay for nothing. The third card turns over once more than it needs to and lands crooked before it straightens. The line under the cards ends on You need exactly one to start."><figcaption>Pick your first engine by what you already pay for: Claude Code with a Claude subscription, Codex with a ChatGPT subscription, Antigravity if you pay for nothing. Tap a card to flip it back.</figcaption></figure>-## Step 2: Install your engine, and log in+Each of the other engines has its own sign in, and the app can install them for you. Pick one only if you already use it.-**This is the only step where you open a terminal.** Once it is done you will not need it again.+| Engine | Command | Installs with |+| --- | --- | --- |+| Gemini CLI | `gemini` | npm |+| Qwen | `qwen` | npm |+| Crush | `crush` | npm |+| Pi | `pi` | npm |+| Copilot | `copilot` | npm |+| Grok | `grok` | its own installer |+| Kimi Code | `kimi` | its own installer |+| Cursor | `cursor-agent` | its own installer |-The terminal is a plain text window for typing commands. On macOS press `Cmd+Space`, type-`Terminal`, press Enter. On Windows press the Start button, type `PowerShell`, press Enter. On Linux-you already know.+You can add more engines later and run several at once, with different agents on different engines. The [Gemini CLI install guide](/blog/how-to-install-gemini-cli/) has that engine's steps.-Find your engine below, copy the block, paste it, press Enter. Wait for it to finish before typing-the next line.+## How do you install an engine and sign in?-{% img "shot-terminal", "Checking that the engines installed. This is the only time you need this window." %}+Open a terminal and paste the lines for your engine one at a time, pressing Enter after each. A terminal is a plain text window for typing commands. On macOS press `Cmd+Space`, type `Terminal` and press Enter. On Windows press Start, type `PowerShell` and press Enter.-### First: do you have Node.js?--Most of these engines install through `npm`, which comes with Node.js. Check with:+**Claude Code** (a Claude subscription). macOS and Linux:```bash-node --version-```--If that prints a version number, you are ready. If it says "command not found", get the **LTS**-build from [nodejs.org/en/download](https://nodejs.org/en/download), install it, then close the-terminal window and open a fresh one.--Antigravity, Claude Code and Cursor can be installed without Node.js. Everything else needs it.--### Antigravity (free, no subscription)--Google's coding agent. The command is `agy`. Homepage:-[antigravity.google](https://antigravity.google/) · Install docs:-[antigravity.google/docs/cli/install](https://antigravity.google/docs/cli/install/)--macOS and Linux:--```bash-curl -fsSL https://antigravity.google/cli/install.sh | bash+curl -fsSL https://claude.ai/install.sh | bash```Windows PowerShell:```powershell-irm https://antigravity.google/cli/install.ps1 | iex-```--Then sign in:--```bash-agy+irm https://claude.ai/install.ps1 | iex```-Your browser opens. Sign in with a Google account.+On any system, close the terminal and open a new one, then run `claude` once and sign in when the browser opens.-### OpenCode (free, bring your own model)--Open source, works with free and local models. Docs:-[opencode.ai/docs](https://opencode.ai/docs)+**Codex** (a ChatGPT subscription). macOS and Linux:```bash-npm install -g opencode-ai@latest-opencode auth login+curl -fsSL https://chatgpt.com/codex/install.sh | sh```-### Codex (with a ChatGPT subscription)--Covered by ChatGPT Plus or Pro. Docs:-[github.com/openai/codex](https://github.com/openai/codex)+Windows PowerShell:-```bash-npm install -g @openai/codex-codex+```powershell+powershell -ExecutionPolicy ByPass -c "irm https://chatgpt.com/codex/install.ps1 | iex"```-On first run it opens a browser to sign you into your ChatGPT account.+Codex can also install through npm, which comes with Node: `npm install -g @openai/codex`. Check Node with `node --version`; the app wants version 20 or newer. If you see an error such as "command not found" instead of a number, you have no Node: get the LTS build from [nodejs.org](https://nodejs.org/en/download), then close the terminal and open a new one.-### Claude Code (with a Claude subscription)+Then close the terminal, open a new one and run `codex`. The first `codex` opens a sign in; choose to sign in with ChatGPT. [How to install Codex CLI](/blog/how-to-install-codex-cli/) covers Windows and updates.-Covered by Claude Pro or Max. Docs:-[docs.claude.com/en/docs/claude-code](https://docs.claude.com/en/docs/claude-code)+**Antigravity** (no subscription). macOS and Linux:```bash-npm install -g @anthropic-ai/claude-code-claude-```--If you would rather not install Node.js, use the standalone installer instead.--macOS and Linux:--```bash-curl -fsSL https://claude.ai/install.sh | bash+curl -fsSL https://antigravity.google/cli/install.sh | bash```Windows PowerShell:```powershell-irm https://claude.ai/install.ps1 | iex+irm https://antigravity.google/cli/install.ps1 | iex```-Then run `claude` and sign in when the browser opens.+Then close the terminal, open a new one, run `agy` and sign in with a Google account.-### The other engines--Same idea: install, then run the command once on its own to sign in.--| Engine | Install it with | Then run | Official docs |-|---|---|---|---|-| Gemini CLI | `npm install -g @google/gemini-cli` | `gemini` | [google-gemini/gemini-cli](https://github.com/google-gemini/gemini-cli) |-| Grok | `npm install -g @xai-official/grok` | `grok` | [docs.x.ai](https://docs.x.ai/build/overview) |-| Kimi Code | `npm install -g @moonshot-ai/kimi-code` | `kimi` | [kimi.com/code](https://www.kimi.com/code/docs/en/kimi-code-cli/guides/getting-started.html) |-| Qwen | `npm install -g @qwen-code/qwen-code@latest` | `qwen` | [QwenLM/qwen-code](https://github.com/QwenLM/qwen-code) |-| Crush | `npm install -g @charmland/crush` | `crush` | [charmbracelet/crush](https://github.com/charmbracelet/crush) |-| Copilot | `npm install -g @github/copilot` | `copilot` | [docs.github.com](https://docs.github.com/copilot/concepts/agents/about-copilot-cli) |-| Cursor | `curl https://cursor.com/install -fsS \| bash` | `cursor-agent` | [cursor.com/docs](https://cursor.com/docs/cli/overview) |-| pi | `npm install -g --ignore-scripts @earendil-works/pi-coding-agent` | `pi` | [pi.dev/docs](https://pi.dev/docs/latest) |--On Windows, the Cursor line is `irm 'https://cursor.com/install?win32=true' | iex` in PowerShell.--### Check it actually installed--Run this, swapping `claude` for whichever command you installed:+**OpenCode** (a key from an AI provider you choose; you pay that provider). The second line asks which provider and takes the key:```bash-claude --version+npm install -g opencode-ai@latest+opencode auth login```-A version number means you are done with the terminal. "Command not found" means the install did-not finish, or the terminal is still running from before the install: close the window, open a new-one, and try again.--<div class="callout warn"><span class="ic">The one thing people skip</span><p><strong>You must-actually sign in.</strong> Installing the tool is not the same as logging into it. Run the command-once on its own and complete the sign-in before moving on, or Munder Difflin will find the engine-and every agent will still fail the moment it starts.</p></div>--## Step 3: Install Munder Difflin--Go to [munderdiffl.in](https://munderdiffl.in) and download the build for your system, or take it-straight from the-[latest release on GitHub](https://github.com/chaitanyagiri/munder-difflin/releases/latest).--{% img "shot-download", "The download page. Pick the file that matches your computer." %}--**macOS.** Download the universal `.dmg`, which runs on Apple Silicon and Intel. Open it and drag-Munder Difflin into Applications. The app is signed and notarized by Apple, so the first launch only-shows the usual question about opening an app downloaded from the internet. Click Open.--**Windows.** Download the setup `.exe` and run it. If Windows SmartScreen shows a warning, click-More info, then Run anyway. If the installer will not start at all, download the portable `.exe`-from the release instead, which needs no installation.--**Linux.** Download the `.AppImage`, make it executable (`chmod +x` on the file) and run it.--**Want to be sure the file is genuine?** Every release carries a `SHA256SUMS.txt`. Run-`shasum -a 256` on the file you downloaded (or `Get-FileHash` in Windows PowerShell) and check that-the value matches that file's line.--## Step 4: Onboarding, six screens--Open the app. It walks you through setup. There are two intro screens and then four numbered steps.--### Screen 1: Who are you?--It asks whether you are technical or not.--**Pick the non-technical option.** This is not a downgrade. It changes the language of the entire-rest of the app, replacing engine ids, flags and model slugs with plain descriptions. You can change-it later in Settings.--{% img "shot-persona", "The first onboarding screen. Choosing the non-technical path rewrites the rest of the setup in plain language." %}--### Screen 2: Meet your office+To check any of them, ask for its version, for example `claude --version`. A version number means you are done with the terminal.-An explanation of what you are about to set up. Read it and continue.+<figure class="mg" data-scene="terminal"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/terminal.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A light terminal window on white, above a row of keyboard keys with a large Enter key. A line types itself: npm install -g @openai/codex. Enter is pressed and a progress bar fills, slips back one block near the end, then finishes with a tick. A second line types codex, a dotted arrow runs from it to a small browser window that slides in over the terminal showing Sign in, and the button in that window turns into a tick. A third line types codex --version and prints three blocks that stand in for a version number, a tick and the words a version number."><figcaption>The whole terminal part of this guide: install the engine, run it once to sign in, then ask for its version to check. The drawing uses Codex; every engine follows the same three steps, though the install command differs. Tap the terminal to run it again.</figcaption></figure>-{% img "shot-welcome", "The office tour screen." %}+<div class="callout warn"><span class="ic">The step people skip</span><p><strong>Installing an engine is not the same as signing in to it.</strong> Run its command once and finish the sign in. Otherwise the agent cannot start. For Claude Code, Codex and Cursor the app shows <strong>Sign in needed</strong>.</p></div>-### Step 1 of 4: A home for the app+### Or let the app install it-The app needs an empty folder to keep its own files: its settings, and your agents' memory.+You can skip the terminal. Setup marks every engine **Installed**, **Installs on first start** or **Not installed**. When you start an agent whose engine is missing, the app says so and offers an **Install** button. Nothing installs until you press it.-Something like `HarnessAgents` in your home folder is fine. It will create it if it does not exist.-This is not where your work goes. It is the app's own filing cabinet.+The app uses the first route that can work on your computer: npm; Node and npm together if Node is missing; the engine's own installer; else a manual route with a link to the docs. The sign in is still yours to finish, in that agent's terminal.-{% img "shot-home", "Choosing a home folder. Make a new empty one." %}+## How do you download and install Munder Difflin?-### Step 2 of 4: Your clone+Get the file for your system from the [download page](https://harnessmd.com/download).-Name your clone and pick the engine that powers it. This is where the engine you installed in Step 2-shows up. If it is not listed as installed, the sign-in did not finish, so go back and run the-command once more.+<figure class="mg" data-scene="download"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/download.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A download page on white shows the address harnessmd.com/download and three buttons labelled macOS, Windows and Linux. A file hangs from a clip under each button: universal .dmg, setup .exe with portable .exe beside it, and .AppImage. Each button is pressed in turn and lets go of its file, which opens a parachute and floats down to a landing pad on the ground. The portable .exe has a carry handle and no parachute, so it drops and bounces."><figcaption>One file per system: a universal .dmg for macOS, a setup .exe or a portable .exe for Windows, and an .AppImage for Linux. Tap a button to drop its file again.</figcaption></figure>-Give your clone the most capable model available to you. It is doing the thinking and delegating,-so it benefits from the extra capability more than the workers do.+* **macOS.** One universal `.dmg` for Apple Silicon and Intel, signed and notarized by Apple. Open it and drag Munder Difflin into Applications. If macOS still refuses to open it, try to open it once, then open **System Settings**, then **Privacy & Security**, press **Open Anyway** within the hour and enter your password.+* **Windows.** Run the setup `.exe`. It installs for your user only and lets you pick the folder. There is also a portable `.exe` that needs no installing.+* **Linux.** Mark the `.AppImage` as executable (`chmod +x` on the file), then run it.-{% img "shot-engine", "Naming your clone and picking its engine." %}+Optional: to confirm the download arrived intact, compare its checksum with `SHA256SUMS-v0.5.6.txt`, published with the release. Run `shasum -a 256` on your download on macOS, `sha256sum` on Linux, or `Get-FileHash` in PowerShell, and compare the value with that file's line.-### Step 3 of 4: Your projects+## What happens the first time you open it?-Add the folders you want agents working in. A project is just a folder. It can hold code, documents,-notes, or anything else.+Setup has four steps. The words in bold are the words on the screen.-You can skip this and add projects later. If you are adding one now, make it a new empty folder.+1. **How are you using Munder Difflin?** Pick **On my own**, or **With my team** if someone sent you an invite code (a team code, not the sign in code below), and press **Continue**. Then comes **Sign in to continue**: press **Open browser to sign in**. The browser never opens by itself. Sign in there. The screen usually moves on by itself. If it does not, copy the code the browser shows, switch back to the app, paste it into **Code from the browser** and press **Use this code**. You see **You are signed in**; press **Continue**.+2. **Name your workspace.** The default name is `HarnessAgents`. Everything your agents do lives in this one folder: their work, memory, tasks and history. **Browse** picks where it goes. The name cannot hold a space, and the app offers a fixed one if you type a space. This step also asks whether you are technical or non-technical.+3. **Meet your orchestrator.** Name him and pick his engine. The engine you pick shows **Installed**, **Installs on first start** or **Not installed**.+4. **A last look at him.** Press **Open the workspace**, and you are in.-{% img "shot-repos", "Adding a project. A project is simply a folder." %}+<figure class="mg" data-scene="signin"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/signin.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two windows on white. An app window on the left holds a button reading Open browser to sign in, an empty field labelled Code from the browser and a pale button reading Use this code. A dotted outline on the right marks the browser, which is not open yet. The first button is pressed and a browser window opens in the outline, with a crowded row of tabs, a tick, the words copy the code and a code drawn as a row of boxes with a dot in each. A copy of the code lifts out of the browser, travels across and drops into the field. Use this code turns yellow and is pressed, and the app window shows You are signed in."><figcaption>Sign in takes two windows: press Open browser to sign in and sign in there. If the screen does not move on by itself, copy the code from the browser, paste it into Code from the browser and press Use this code. Tap the app window to watch it again.</figcaption></figure>-### Step 4 of 4: Permissions+Then comes the screen that stops people: the Pro sales screen, **Run your agents like a company**, with three buttons.-The important screen. It asks how much agents can do on their own.+* **Continue with the free version** opens the Classic view. This is the one you want if you have no licence.+* **I have a licence key** lets you paste a key.+* **Get PRO** opens the checkout in your browser.-**On** means agents carry out tasks without stopping to ask. Smoothest, and how you will eventually-want it.+<figure class="mg" data-scene="fork"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/fork.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A mini golf course on white, seen from above. A putter taps a ball off a mat marked after setup. The green lane splits into three branches, and a button sits on each one: Continue with the free version, I have a licence key and Get PRO. Each branch ends at a hole, marked Classic view, paste a key and checkout in your browser. The ball rolls to the split, stops, leans toward two of the branches, then takes the one under Continue with the free version, which turns yellow. It circles the rim of the hole once before it drops in, and a tick appears on the hole beside Classic view. The other two buttons then light up in turn. Tapping a branch sends the ball down that one."><figcaption>The first screen after setup offers Continue with the free version, I have a licence key and Get PRO. Without a licence, press Continue with the free version to reach the Classic view. Tap a branch to take it.</figcaption></figure>-**Off** means agents pause and ask you before changing files or running commands.+Classic is the free view: the office floor. The Pro view adds one window with screens for Tasks, Inbox, Automations, Memory and Capabilities, plus the Stapler, a small window above every app. You can move between the two later with the switch in the title bar, **Classic** and **PRO**. Without a licence the second one reads **TRY PRO**, and pressing it brings back this screen.-**Start with off.** Watch a few tasks run, see what the agents actually try to do, and turn it on-when you are comfortable. You can change this any time.+## What do you do in your first ten minutes?-{% img "shot-permissions", "The autonomy choice. Start with agents asking permission." %}--## Step 5: Your first job--You are looking at the office floor. Your clone is there. Nobody else yet.--Talk to your clone the way you would talk to a capable new hire who has not met you yet. Be specific-about the outcome, and do not bother specifying the steps.+You are in the office. Your orchestrator is there, and nobody else yet. Talk to him the way you would brief a capable new hire: the outcome, the folder, and what done looks like.Weak: *"organise my files"*-Better: *"Look in my Documents/receipts folder. Sort every PDF into subfolders by year, based on the-date inside the document, not the filename. Tell me about anything you cannot read."*--Your clone will decide it needs a worker, hire one, and hand the job over. You will see a new-character appear at a desk.--{% img "shot-floor", "The office floor on a fresh setup. Michael, your clone, is the only one here yet. The Command Center on the right is where work gets dispatched." %}--Click any agent to open its terminal and watch it work in real time.--### Things people actually use this for--Not all of these are coding.--- **Documents.** Sort a folder of files, rename them consistently, pull the numbers out of fifty- invoices into one spreadsheet.-- **Research.** Give an agent a question and a place to write, and let it work through it while you- do something else.-- **Code.** The obvious one. Fix a bug, add a feature, upgrade a dependency, write the tests.-- **Repeating chores.** Anything you do every Monday, handed to a schedule instead.-- **Long jobs.** The ones you keep not starting because they take four hours. Start one before bed.--## Step 6: Add agents yourself--Your clone hires people on its own, but you can hire directly too.--Click add agent. You pick a name, an engine, a model and a project. Different agents can run on-different engines, so you can put a free engine on the boring jobs and your paid one on the hard-ones.--{% img "shot-addagent", "The add agent panel: identity, workspace, engine, briefing. Yes, you pick which Office character they are." %}--Agents persist. Close an agent and its memory and inbox reattach when you bring it back, with the-same identity.--## Step 7: Work that starts without you--Open the triggers tab. Everything here starts work when you are not typing.--{% img "shot-triggers", "The triggers tab: schedules, context, webhooks and organisation." %}--**Schedules** run a prompt on a repeating clock. You give it a label, choose who it goes to, and-write the prompt, which is sent word for word on every run. This is the feature that turns the app-from a tool you use into something that runs on its own.--Good first schedule: every morning, ask an agent to check something and write you a short summary.--{% img "shot-schedule", "Creating a schedule. The prompt is sent word for word on every run." %}--**Context** handles what happens as an agent's memory fills up (see below).--**Webhooks** let an outside system post work in.+Better: *"Look in my Documents/receipts folder. Sort every PDF into folders by year, using the date inside the document, not the file name. Tell me about any file you cannot read."*-**Organisation** lets a teammate's Munder Difflin send work to yours.+Your orchestrator decides whether he needs help, and hires it. A new agent appears at a desk. When you hire one yourself, you pick its face from fifteen built in avatars. Click any agent to open its terminal and watch it work.-## Step 8: Context and auto-compaction, briefly+When that works, [your first hour](/blog/your-first-hour-with-munder-difflin/) covers teams, schedules, Slack and the Pro screens. For ideas, [every agent on our own floor](/blog/ai-agent-use-cases-our-office/) shows the jobs we give ours.-Every AI model has a limited working memory, called its context. A long-running agent will fill it.-When it is full, the agent gets confused or stops.+## What is the IDE button for?-**Auto-compaction** is the fix. When an agent's context gets close to full, the app has it summarise-what has happened so far and carry the summary forward instead of the raw history. The agent keeps-going without losing the thread.+Click an agent and press **IDE** in its panel. The button opens a code editor on that agent's working folder, inside the app. Use it to see what an agent just did without leaving the office.-You control this under Context in the triggers tab: compact automatically, clear entirely, or leave-it alone. The default is sensible and most people never touch it.+* Open, edit and save any file, or make a **New file**.+* See what the agent changed: **Changes**, **History** and **Compare**.+* Preview Markdown files, the plain text format agents write reports in, as **code**, **split** or **preview**.+* Open images.-## Step 9: Settings worth knowing+It is the quickest way to check an agent's work. Ask for a report, open the IDE and read the report in preview next to the files it describes.-Seven sections. These are the ones that matter early:+<figure class="mg" data-scene="ide"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/ide.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A light code editor window on white. Its title bar holds a button reading IDE and the words an agent's working folder. Under it are three tabs, Changes, History and Compare, and a switch reading code, split and preview. A file list runs down the left side with a New file button at the bottom, and one file is selected. The code pane shows the file as grey bars, one bar per line. Two new lines type themselves in at the bottom, and a yellow dot marks the selected file as changed. The Changes tab lights up and the two new lines are marked in green. The switch then slides to split, and a formatted preview with a heading and a bullet list opens beside the code. It slides to preview, where the preview fills the pane, and back to split. The scroll bar handle is tiny, because the file is long."><figcaption>The IDE button opens an editor on an agent's working folder: edit and save files, see Changes, History and Compare, and preview Markdown as code, split or preview. Tap the switch to change the view.</figcaption></figure>-- **General.** Change your home folder, switch between technical and plain language, pick your- language. The app runs in English, Simplified Chinese and Arabic.-- **Prerequisites.** Checks whether your engines are actually installed and logged in. **This is the- first place to look when something is not working.**-- **Agents & Models.** Which engines are available and which models each uses. New models show up- without an app update, because the model pickers read the latest list when the app launches.-- **Autonomy & Budgets.** The permission choice from onboarding, plus spending limits per agent.- Worth setting a cap before leaving anything running overnight.-- **Connections.** Slack, webhooks, MCP and the REST API.-- **Voice.** Talk to your agents out loud.-- **Memory & Knowledge.** What your agents remember between sessions.+## What can you add after setup?-{% img "shot-settings", "The settings panel. Prerequisites is where you check that an engine is really installed." %}+Almost everything. Setup asks for the minimum, and the rest lives in **Settings**.-## Step 10: Focus mode+* **Prerequisites** shows whether each engine and helper tool is installed.+* **Agents & Models** is where a second or third engine comes in.+* **Autonomy & Budgets** holds how much agents may do alone, their budgets, and whether temps (short lived agents for one job) may be started.+* **Connections** is Slack and the other ways work arrives from outside. The [Slack guide](/blog/connect-slack-to-munder-difflin/) walks through it.+* **Memory & Knowledge** covers what agents remember. Search by meaning needs `uv` and MemPalace.+* **Keys & Secrets** is where saved keys live. OpenCode, Crush, Pi and Qwen take a key; Claude Code and Codex use their own sign in.-When you want to stop watching the office and work with one agent, press the fullscreen control on-its terminal. The floor disappears and you get a full-width terminal.+## What do you do when an install fails?-Press `Esc` to come back.+Read the message first. The app says what is wrong in a sentence, and most screens carry the button that fixes it.-{% img "shot-focus", "Focus mode. One agent, full width, no office." %}+| What you see | What to do |+| --- | --- |+| "… is not installed on this computer.", with your engine's name in front | Press **Install**. If it stops, **Try again**. **Open the docs** and **Set up manually** are the fallbacks. |+| **Sign in needed** | Finish the engine's sign in in that agent's terminal, then press **Setup complete, start agent**. |+| "Signing in needs the internet. Your work is safe on this machine. Reconnect and try again." | Reconnect, then try again. Nothing is lost. |+| "Munder Difflin could not verify the secure connection to our server." | Check the computer's date and time. If they are right, ask whoever runs your network or security software to allow the address in the message, then press **Try again**. |+| "A folder name cannot hold a space." | Press the button that reads **Use** and the suggested name. |+| A note that a port (a numbered door programs listen on) was in use and the session listens on another one | Nothing, unless something on this computer calls that port, such as a webhook. Then point it at the new number. |+| macOS will not open the app | Try to open it once, then **System Settings**, then **Privacy & Security**, then **Open Anyway**. |-## When something goes wrong+If the message does not settle it, work up this list and stop at the first step that fixes it. This order is our advice from running the app every day; the app itself only offers the first step.-**An agent dies the moment it starts.** Almost always the engine is installed but not logged in.-Open Settings, then Prerequisites. Or run the engine's command once in a terminal and finish the-sign-in.+1. **Check again.** Press **Check again** or **Try again** on the screen.+2. **Restart the agent.** Close that agent and start it again.+3. **Reopen the app.** Quit Munder Difflin completely and open it again.+4. **Restart the computer.** The last resort, if step 3 did not help.-**"Engine not installed" during onboarding, but you installed it.** The app looks for the command on-your system path. Close the app completely and reopen it, since it reads your environment at launch.+<figure class="mg" data-scene="fixes"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/fixes.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Four stepping stones cross a small stream on white, between two green banks, each stone a little bigger than the last. They are labelled Check again, restart the agent, reopen the app and restart the computer. Every stone carries the same round restart arrow, drawn around a button, a terminal, an app window and a computer. Above the near bank a crooked card reading Install shows a progress bar snapped in two and a yellow warning mark. A pebble hops from the bank onto the first stone, the arrow spins, the broken half of the bar lifts and drops again, and the stone gets a lilac cross. The second stone goes the same way. On the third stone the bar mends and fills, the card straightens, a tick appears and the pebble stays there. The fourth stone is never needed. Past the far bank a speech bubble reads Still stuck? Ask your orchestrator. Tapping a stone makes that stone the one that works."><figcaption>When an install fails, work up four steps and stop at the first that fixes it: Check again, restart the agent, reopen the app, restart the computer. Tap a stone to hop that far.</figcaption></figure>-**Windows blocks the installer.** If SmartScreen flags it, choose More info, then Run anyway, or use-the portable build.+Still stuck? Ask your orchestrator: *"The Codex install failed. Read the log and tell me why."* If the engine that failed is the one your orchestrator runs on, there is nobody to ask yet: run that engine's lines from the terminal section above, then press **Check again**. The log is `hive/log.jsonl` inside your workspace folder, and a crashed agent's raw output is in `hive/crashes` beside it.-**macOS asks whether to open an app downloaded from the internet.** That is normal for anything you-download. Click Open.+Three things look like failures and are not:-**Nothing is happening.** Check whether message delivery is paused in the Command Center. Paused-means queued work is being held for every agent, and nothing is lost when you switch it back on.+* **An agent reads "looping".** That is the circuit breaker warning an agent that repeats itself or spends fast. It is the app working.+* **An agent's session was cleared.** By default the app clears a session when the agent's work switches and the session has run for 60 minutes, to save tokens. A file called `handoff.md` in the agent's folder carries the thread across.+* **Everything is slow with many agents.** A computer with 8 GB of memory struggles with a large team. Run fewer at once.-## The short version+## What is the short version?-1. Pick an engine. Antigravity if you want free, Codex if you pay for ChatGPT, Claude Code if you- pay for Claude.-2. Install it and sign in. One terminal command. The only one.-3. Download Munder Difflin and install it.-4. Six onboarding screens. Pick the non-technical path, use a new empty folder, start with agents- asking permission.-5. Give your clone a real job and watch what happens.+Seven steps, about twenty minutes.-{% img "note-2" %}+1. Pick one engine: Claude Code, Codex, or a free one.+2. Download Munder Difflin and open it.+3. Press **Open browser to sign in** and sign in.+4. Name your workspace and your orchestrator.+5. Press **Continue with the free version**.+6. If the engine is missing, press **Install**, then sign in to it.+7. Give your orchestrator a real job in a new empty folder, and watch.-The app is [open source on GitHub](https://github.com/chaitanyagiri/munder-difflin). If you get-stuck, the [Discord](https://munderdiffl.in) is the fastest place to ask.+<link rel="stylesheet" href="/blog/assets/media/how-to-install-and-use-munder-difflin/motion.css"><script defer src="/blog/assets/media/how-to-install-and-use-munder-difflin/motion.js"></script>diff --git a/blog/src/posts/omniagent-alternatives.md b/blog/src/posts/omniagent-alternatives.mdindex 4016a447..88d41312 100644--- a/blog/src/posts/omniagent-alternatives.md+++ b/blog/src/posts/omniagent-alternatives.md@@ -6,7 +6,7 @@ category: comparisonscategoryLabel: Comparisonstype: Technicalpinned: true-pinOrder: 7+pinOrder: 8primaryKeyword: "omniagent alternatives"secondaryKeywords: ["omnigent alternatives", "omnigent databricks", "omnigent meta harness", "omniagent harness", "omnigent vs opencode"]tags: ["Comparisons", "Multi-Agent", "Claude Code", "Tools", "Open Source"]diff --git a/blog/src/posts/your-first-hour-with-munder-difflin.md b/blog/src/posts/your-first-hour-with-munder-difflin.mdindex 7a7f0c31..ddaf62bd 100644--- a/blog/src/posts/your-first-hour-with-munder-difflin.md+++ b/blog/src/posts/your-first-hour-with-munder-difflin.md@@ -1,289 +1,174 @@----title: "Your First Hour With Munder Difflin: Onboarding to a Pro Office"-description: "Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email."+title: "Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office"+seoTitle: "Munder Difflin Onboarding: Your First Hour With 0.5.6"+description: "Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens."date: 2026-07-03-updated: 2026-09-14+updated: 2026-10-14category: guidescategoryLabel: Guidestype: Non-technicalpinned: truepinOrder: 2primaryKeyword: "munder difflin onboarding"-secondaryKeywords: ["getting started with munder difflin", "munder difflin tutorial", "munder difflin pro", "munder difflin setup", "scheduled agent automations", "munder difflin stapler"]-tags: ["Guides", "Onboarding", "Getting Started", "Multi-Agent", "Local-First"]+secondaryKeywords: ["getting started with munder difflin", "munder difflin tutorial", "munder difflin pro", "munder difflin temps", "munder difflin automations", "munder difflin stapler", "munder difflin first hour"]+tags: ["Guides", "Onboarding", "Getting Started", "Multi-Agent"]author:name: Chaitanya Giriinitials: CGfaq:- q: "How long does it take to get Munder Difflin running?"- a: "About ten minutes if Claude Code, Codex or OpenCode is already installed and signed in: download, sign in, answer four setup steps. Allow the rest of the hour to hire a small team and set up your first automations."- - q: "Do I need an account to use Munder Difflin?"- a: "Yes, in 0.5.2. The first launch asks you to sign in with an email address in your browser, on the free plan as well. Pro then needs a licence key on top of that account."- - q: "What does Pro add over the free version?"- a: "The free version is the classic office. Pro adds a single window workspace with screens for the orchestrator, agents, agent rooms, tasks, inbox, automations, memory, capabilities and temps, plus the Stapler. It runs on one machine at a time; the pricing page has the rest."- - q: "Can one Pro office mix Claude Code, Codex and OpenCode agents?"- a: "Yes. Each agent picks its own provider and model when you add it, and can have its own token cap. Claude Code and Codex use their own logins; OpenCode can use a provider key you save under Settings."+ a: "Setup is four short steps when an AI engine is already installed and signed in: download, sign in, name your workspace and your orchestrator. The rest of the hour goes on a first job, a small team and the first work that starts without you."+ - q: "Do I need Pro for this first hour?"+ a: "No. Agents, schedules, the context rules, webhooks, Slack and temps are in both views. The Pro view adds one window with screens for Tasks, Inbox, Automations, Memory and Capabilities, plus the Stapler, a small window above every app."+ - q: "What is a temp?"+ a: "A temp is a short lived agent started for one job. The orchestrator or any lasting agent can start one, and the app can start one to answer an inbound message. It does the job, reports done and leaves."+ - q: "Can one office mix Claude Code, Codex and other engines?"+ a: "Yes. Each agent picks its own engine and model when you add it, and can have its own token cap. Claude Code and Codex use their own sign in; OpenCode, Crush, Pi and Qwen take a key."- q: "Does the office keep working when I close my laptop?"- a: "No. Agents, automations, Slack and webhooks all run on your machine, so they stop when the app closes or the laptop sleeps. A schedule missed while the laptop slept runs once when the app wakes; a weekly one only if that is within six hours of its slot."- - q: "Is the Stapler free?"- a: "No. The Stapler is Pro only, and it is off until you turn on **Show the stapler** on the Stapler screen."+ a: "No. Agents, schedules, Slack and webhooks all run on your own computer, so they stop when the app closes or the laptop sleeps."+ - q: "Why did my agent's session clear itself?"+ a: "That is a default in 0.5.6 that saves tokens. When an agent's work switches and its session has run for 60 minutes, the app clears the session, and a handoff file in the agent's folder carries the thread across."startHere: truestartOrder: 2----Your first hour with [Munder Difflin](https://harnessmd.com/download) 0.5.2 goes like this. Download the app, install one coding CLI, sign in and name your orchestrator. Then activate Pro and set the office up to work without you: a mixed team of agents, a pull request reviewer, a Slack responder, a morning email brief and a content agent on a schedule.+Your first hour of Munder Difflin onboarding goes like this: sign in, give your orchestrator one real job, hire a small team, set up work that starts without you, then learn the Pro screens. This guide is written for version 0.5.6 and follows the clock.-The app screenshots below are the real 0.5.2 Pro app, with personal details blurred. The order follows a new user from the download page to a working office.+[Munder Difflin](https://harnessmd.com/download) is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer. Not installed yet? The [install guide](/blog/how-to-install-and-use-munder-difflin/) covers the download, the engines and every fix. For ideas on what to hand your agents, [every agent on our own floor](/blog/ai-agent-use-cases-our-office/) shows what ours do.-## Where do you download Munder Difflin?+<figure class="mg" data-scene="hour"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/hour.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A bus route map on white: one road runs left to right, bends round a hairpin past a pond and comes back, with six stops on it. Each stop has a minute mark and a name: Minute 0, sign in; Minute 10, a first job; Minute 20, a small team; Minute 35, work that starts without you; Minute 45, temps and tokens; Minute 50, the Pro screens. A small yellow bus drives from stop to stop, the road turns blue behind it, and each stop lights up yellow as the bus arrives. The bus leans through the hairpin and leaves two skid marks."><figcaption>The first hour as a route: sign in, a first job, a small team, work that starts without you, temps and tokens, then the Pro screens. Tap a stop to send the bus there.</figcaption></figure>-Download it from [harnessmd.com/download](https://harnessmd.com/download). You get there from **Download free** on [munderdiffl.in](https://munderdiffl.in/), or from **Download** in its top bar.+## Minute 0: what do you need, and how do you sign in?-{% img "a1-landing" %}+You need the app, one AI engine and an email address. The app needs nothing else installed, and it can install a missing engine for you later.-On 14 Sep 2026 the page listed version 0.5.2, published 9 Sep 2026, in three builds:+The first launch asks **How are you using Munder Difflin?** Pick **On my own**, then:-* **macOS:** one build for Apple silicon and Intel, 246 MB.-* **Windows:** x64 installer, 132 MB.-* **Linux:** x86_64 AppImage, 175 MB.+1. Press **Open browser to sign in**. The browser does not open by itself.+2. Sign in in the browser. The screen moves on by itself; if it does not, paste the code into **Code from the browser** and press **Use this code**.+3. **Name your workspace**, **Meet your orchestrator**, then **Open the workspace**.-{% img "a3-download" %}+Next comes the Pro sales screen, **Run your agents like a company**. Press **Continue with the free version** to reach the Classic view, or **I have a licence key** to turn on Pro. The steps through minute 45 work in both views; a few labels below are the Pro view's.-Open the file the usual way for your system: drag the app to Applications on a Mac, run the installer on Windows, or make the AppImage executable on Linux. [What changed in this version](https://harnessmd.com/download#whats-new) is written out on the download page.+## Minute 10: how do you give your orchestrator a first job?-## What do you need installed before the first launch?+Brief him like a capable new hire: the outcome, the folder, and what done looks like. Your orchestrator is the one agent you talk to. He is called Michael until you rename him.-You need at least one coding CLI on your path, signed in with your own subscription or key. Munder Difflin supports twelve; this guide uses Claude Code, Codex and OpenCode. The commands below come from each vendor's install docs, checked on 14 Sep 2026.+A vague brief gets a vague job. Compare:-### Claude Code+* Weak: *"clean up my project"*+* Better: *"In the folder notes-app, find every function with no test. Write the tests, run them, and give me a list of what failed and why. Do not change any function."*-```bash-# macOS, Linux, WSL-curl -fsSL https://claude.ai/install.sh | bash-```+Your orchestrator turns the brief into tasks and works them, hands them to an agent you added, or starts a temp. Once there is one, click any agent to watch its terminal. To read what it wrote, press **IDE**: an editor opens on the agent's folder, with **Changes** to show what was edited and a Markdown **preview** for reports.-```powershell-# Windows PowerShell-irm https://claude.ai/install.ps1 | iex-```+<figure class="mg" data-scene="brief"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/brief.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two notepads on white above a long toaster marked your orchestrator. The pad tagged Weak reads clean up my project; the pad tagged Better reads the outcome, the folder, what done looks like. A slip drops from the weak pad into the first slot, the toaster works, and one grey card with a question mark pops up, wobbles and keeps smoking a little. A slip drops from the better pad into the second slot, and three task cards pop up, hop apart and each one gets a tick in its box."><figcaption>A brief with the outcome, the folder and what done looks like turns into clear tasks. A vague one turns into a question. Tap a note to drop it again.</figcaption></figure>-Homebrew (`brew install --cask claude-code`) and WinGet (`winget install Anthropic.ClaudeCode`) work too. Run `claude` once in any folder and sign in with your Claude account.+## Minute 20: how do you hire a small team?-### Codex+Press **Add an agent**. Pick a face, an engine, a model and the folder it works in. Each agent can run on a different engine, and that is the point: put your best model on the hard work and a cheaper one on the chores.-```bash-# macOS, Linux-curl -fsSL https://chatgpt.com/codex/install.sh | sh-# or, with Node installed-npm install -g @openai/codex-```+One split to try on day one:-Run `codex` and sign in with ChatGPT or an API key. Watch out for `npm install -g codex` without the `@openai/` scope: that is an unrelated package. [How to install Codex CLI](/blog/how-to-install-codex-cli/) covers Windows and updates, and the [Munder Difflin install guide](/blog/how-to-install-and-use-munder-difflin/) has the commands for all twelve CLIs.+* a builder on **Claude Code**, writing the feature+* a reviewer on **Codex**, reading every change the builder makes+* a helper on a free engine such as Antigravity, for docs and tidying-### OpenCode+In the Pro view, each agent also has a **Token cap: the session budget before a restart**. Leave it on **Workspace budget** unless one agent needs its own. Sign every engine in before you hire on it, or the new agent cannot start; a Claude Code, Codex or Cursor agent stops at **Sign in needed**.-```bash-# macOS, Linux-curl -fsSL https://opencode.ai/install | bash-# or-npm install -g opencode-ai-```+Our own floor has eleven kinds of seat: an orchestrator, mission control, a community lead, growth, motion graphics, research, a founder's office, a scrum master, an engineering lead, builders, and a designer. You do not need eleven. You need a reviewer.-On Windows, OpenCode's docs list `choco install opencode` and `scoop install opencode`. Start `opencode` and run `/connect` to choose a model provider.+<figure class="mg" data-scene="team"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/team.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three dotted outlines of office chairs stand in a row on white under a yellow button reading Add an agent, with the roles builder, reviewer and helper below them. A pointer presses the button three times; each press rolls in a chair, and an engine tag appears under it: on Claude Code, on Codex, on a free engine. The third chair arrives still spinning. A sheet of paper leaves the builder's chair, crosses to the reviewer's chair, is read line by line, gets a tick and comes back. Then the helper's heap of papers squares itself into a neat stack."><figcaption>A first team of three: a builder on Claude Code, a reviewer on Codex that reads every change, and a helper on a free engine for docs and tidying. Tap the builder to send another change for review.</figcaption></figure>-With Node installed, npm can put all three on your machine in one command. For the screenshot we installed them into a throwaway folder, then checked each with `--version`:+## Minute 35: how do you set up work that starts without you?-{% img "b1-terminal" %}+With an automation. In the Pro view open **Automations** and press **New automation**; in the Classic view the same things live under Triggers. Two kinds start there, a schedule and a webhook, and Slack is a third way in.-If you skip this step, the app has a fallback. When you start an agent whose CLI is missing, 0.5.2 installs it in that agent's own terminal, installing Node first if it has to, then relaunches the agent. You finish the CLI's sign in right there. **Settings, Prerequisites** shows the status of each tool, with a **Check now** button. For memory search you also want `uv` and MemPalace, covered under memory below.+**A schedule** sends a message to an agent on a clock. Choose **Schedule**, then fill in four things: a **Label**, **When** (an **Interval** or **Weekly**), **Sends to**, and the **Message**. The message is what the agent should do when it fires. A good first one: every weekday morning, ask an agent to check something and write you three lines.-{% img "b2-prerequisites" %}+<figure class="mg" data-scene="schedule"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/schedule.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A card on white titled Automations with a yellow New automation button; on the right, a horizon with an empty sky and an office chair. A pointer presses New automation, picks Schedule over Webhook endpoint, and four fields fill in: Label, morning check; When, Weekly rather than Interval; Sends to, an agent; Message, check something and write you three lines. The pointer leaves. A sun climbs over the horizon, the card fires by itself, and a paper slip flies to the chair and gets a tick. The pointer peeks back in to check, then leaves again."><figcaption>A schedule has four parts: a Label, When (an Interval or Weekly), Sends to, and the Message the agent gets every time it fires. Tap the card to fire it again.</figcaption></figure>-## How do you sign in on the first launch?+**A webhook** lets any tool that can send a web request start work. Choose **Webhook endpoint**. A new endpoint starts switched off: choose how much you trust the caller, copy its address and its secret into the tool that will send, then turn it on. The sender must hold the secret.-You sign in through your browser with an email address; 0.5.2 asks for an account even on the free plan. The first screen asks **How are you using Munder Difflin?** Pick **On my own**, or **With my team** if an admin sent you an invite code.+**Slack** lets people send work from a channel and read the answer in the same thread. It is under **Settings**, then **Connections**. In 0.5.6 one bot can listen to **Every channel the bot is in**, and a direct message to the bot is always answered. For direct messages, add `im:history`, `im:read`, `mpim:history` and `mpim:read` on your Slack app's **OAuth & Permissions** page at api.slack.com, then follow the app's own **How to get these** steps for the events. You choose the agent under **Who answers inbound messages**. If Slack fails to start, it now comes back by itself. The [Slack guide](/blog/connect-slack-to-munder-difflin/) has the full setup.-{% img "c1-entry" %}+<figure class="mg" data-scene="slack"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/slack.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a Connections card under Settings with a Slack switch and a box for Every channel the bot is in; below it, three chat cards labelled a channel, another channel and a direct message. A pointer switches Slack on and ticks the box, and an office chair rolls into the empty seat under Who answers inbound messages. A string runs from each card to the chair. Each card posts a message, a note travels down its string to the chair, and a reply comes back and lands under that message as a thread. Then Slack switches itself off, the strings go slack, and it switches itself back on."><figcaption>In 0.5.6 one bot can hear every channel it is in, a direct message is always answered, and the reply comes back in the same thread. Tap a card to send a message.</figcaption></figure>-**On my own** leads to **Sign in to continue**, which opens your browser. If your browser does not come back to the app on its own, the screen lets you paste a code instead. New to Munder Difflin? The same sign in page on [harnessmd.com](https://harnessmd.com/signup) creates the account.+## Minute 45: what are temps, and how do you keep token use down?-{% img "c2-signin" %}+A temp is an agent for one job. It gets an objective, does the work, reports done and leaves. There is no button for one. You ask: *"Michael, start a temp to rename every file in this folder to lower case."*-## How do you set up your workspace and orchestrator?+Three things changed in 0.5.6:-Three short steps follow the sign in, and the header counts all four for you.+* **Any lasting agent can start a temp**, not only the orchestrator. A temp cannot start a temp.+* **A temp can answer inbound messages.** Wherever you pick who answers (Slack, a webhook, a schedule's **Sends to**, the Stapler's **Send captures to**), the list now offers **A temp** and a temp that is a clone of one of your agents.+* **You set the limits.** In the Pro view, on the orchestrator's **Budget & breaker** tab, **Temps per agent** runs from 0 to 4 and **Max temps at once** caps how many temps run in the whole office at the same time. At 0, agents cannot start temps.-1. **Name your workspace.** A workspace name (HarnessAgents by default), a folder for the app's own files, and whether you are technical or non technical. That last answer only changes how much jargon the app shows.-2. **Meet your orchestrator.** Name the one agent that runs the others, then pick its engine and model. Each engine shows **Installed**, **Installs on first start** or **Not installed**. On Claude Code the default model is Opus 4.8 with a 1M context window, which suits the job: the orchestrator does the planning and delegating.-3. **Open the workspace.** A last card shows the orchestrator and your first agent, with **Open the workspace** at the bottom.+<figure class="mg" data-scene="temp"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/temp.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three office chairs stand on white over the words your lasting agents, beside the dotted outline of a folding chair. A message card reading one job waits at the top left, then flies across while a yellow folding chair slides in folded, half opens, snaps shut, then opens properly and gets a tag reading A temp. The message lands on its seat, a done card pops up and flies to the first office chair, and the folding chair folds flat and slides away. The three office chairs never move."><figcaption>A temp answers one message or does one job, reports done and leaves; your lasting agents keep their seats. Tap the message to send another.</figcaption></figure>-We call the orchestrator Michael in this guide. Call yours whatever you like.+Tokens are the other limit. Long sessions fill an agent's context, and a full context is slow and uses up your plan's limits faster. 0.5.6 ships with defaults that deal with it:-{% img "c4-orchestrator" %}+* **Compact when context fills** summarises a long session and carries the summary forward.+* **Clear on a switch of work** clears the session when an agent moves to different work and the session has run for 60 minutes. A file called `handoff.md` in the agent's folder carries the thread across, so the next session knows where the last one stopped.+* **No hourly standup.** A new install no longer pings the orchestrator every hour.-{% img "c5-engines" %}+The first two are rules in **Automations** under **Context** (in the Classic view, the Triggers tab). Leave them on.-## How do you get Pro?+<figure class="mg" data-scene="tokens"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/tokens.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a tall glass on a coaster stands beside a card titled Context under Automations, with two rules switched on: Compact when context fills, and Clear on a switch of work. A pill below the card reads Leave them on. Paper slips drop into the glass until the pile nears a dashed line. The first rule lights up, a press with a yellow knob swings over and pushes down, and the slips squash into one slip labelled summary. A few more slips land on top. Then the second rule lights up: the glass turns upside down and is shaken, the slips fall out, the summary clings on until a second shake, and the glass turns back. A yellow note labelled with the handoff file's name drops into the empty glass and stays."><figcaption>Two defaults in 0.5.6 keep sessions small: Compact when context fills summarises a long session, and Clear on a switch of work clears it and leaves a handoff file to carry the thread. Tap the glass to fill it again.</figcaption></figure>-When you open the workspace without a licence, the app shows **Go PRO on this machine**. You have three choices:+## Minute 50: what does the Pro view add?-* **Continue with the free version** takes you to the classic office, free, with everything local.-* **Get PRO** opens checkout in your browser. Pro runs on one machine at a time. Plans and prices are on the [pricing page](https://munderdiffl.in/#pricing).-* **I have a licence key** opens **Enter your license key**. Paste the key from your console at app.harnessmd.com, press **Activate**, and it is bound to this machine.+One window with a sidebar, and a screen for each part of the office. The switch in the title bar moves between **Classic** and **PRO**; without a licence it reads **TRY PRO**.-{% img "c7-paywall" %}+<figure class="mg" data-scene="sidebar"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/sidebar.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. An app window on white with a switch in the title bar reading Classic and PRO, with PRO selected, and a sidebar of six rows: Tasks, Inbox, Automations, Memory, Capabilities, Stapler. A yellow highlight walks down the rows, and as each one lights, its one line note writes itself into the pane beside it: every card in the office; one conversation per agent; Scheduled, Context and Webhooks; Search every agent's memory; what each agent can reach; a small window that floats above every app. When the highlight reaches Stapler, that row lifts off the sidebar and floats above it with a small shadow under it."><figcaption>The Pro view is one window with a sidebar row for Tasks, Inbox, Automations, Memory, Capabilities and the Stapler. Tap a row to see what it is for.</figcaption></figure>-A key that is already active on another machine is refused: release it in your console, then activate it again. After that the app rechecks the licence in the background, and you do nothing.+{% img "pro-sidebar-056", "The real Pro sidebar: Tasks, Inbox, Automations, Memory, Capabilities and Stapler. From a preview build of 0.5.6 with sample data." %}-{% img "c8-license" %}+**Tasks** is every card in the office, as a **Board**, a **List** or **Cards**. The habit that pays: start the day on the **Asks me** filter, which shows only the cards waiting on you. (It is a filter in Tasks; **Ask me** in the Inbox, below, is where you answer.) You can drag cards between columns. Nothing changes for your team until you send your edits, and then your orchestrator makes the change.-## What does a set up Pro office look like?+**Inbox** holds one conversation per agent, plus **Ask me**, **Slack** and **Webhooks**. **Ask me** is the list of questions waiting on a person. Answer in one line; answering unblocks the task. This is the screen that keeps an office moving while you do something else.-A Pro office is one window. The sidebar lists the orchestrator and your agents, and has a screen each for **Agents**, **Tasks**, **Inbox**, **Automations**, **Memory**, **Capabilities**, **Stapler** and **Temps** (short lived agents the orchestrator starts for one job).+{% img "ask-me-056", "Ask me in an agent's Inbox: each question waits with an Answer button. From a preview build of 0.5.6 with sample data." %}-The **Agents** screen puts the orchestrator on top, with today's spend, the circuit breaker, open tasks and questions waiting on you. Below him, each agent has a card with its engine, model, current ticket and a live terminal preview.+<figure class="mg" data-scene="askme"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/askme.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a large card with an Ask me chip and the word Inbox shows a question as grey lines with a lilac question mark, an answer field reading Answer in one line, a blue send button and a chip reading waiting on you. Beside it, three smaller question cards wait in a column under the words waiting on a person. A short answer types itself in as one line and is sent; the question mark turns into a mint tick and the chip becomes unblocks the task. The card slides away, the next question card grows into its place, and the queue shuffles up; the last card arrives a little too fast and bumps the one ahead. The new question gets its one line answer and its tick too."><figcaption>Ask me collects the questions waiting on a person. A short answer unblocks the task, and the card moves on. Tap the card to ask again.</figcaption></figure>-You give work to the orchestrator in his terminal, the way you would brief a capable new hire: the outcome, the folder, and what done looks like. He turns it into tasks, hands them to agents and brings you only what needs your decision. How he decides is in [how the orchestrator routes work](/blog/how-the-god-orchestrator-works/).+**Automations** is the list you met at minute 35, with filters for **Scheduled**, **Context** and **Webhooks**, Each schedule shows when it **Last fired**, each webhook lists its **Recent** arrivals, and **History** opens the Inbox, where webhook messages are answered.-{% img "d2-agents" %}+**Memory** lets you **Search every agent's memory** at once and see the result as a **Graph** or a **List**. Use it before you brief: if an agent already solved this last week, the note is here. Narrow a search with **Whose memory** and **Kind**.-## What can your office do in its first hour?+**Capabilities** shows what each agent can reach: MCP servers, skills and tools.-Each of these six use cases takes a few minutes to set up.+**The Stapler** is a small window that floats above every app. On the **Stapler** screen, turn on **Show the stapler**, then click it for its ring:-### Use case: a mixed team of agents+* **Screenshot** captures a region of the screen and sends it, with a note, to the agent you pick under **Send captures to**.+* **Record message** takes a spoken message and shows you the transcript before it goes.+* **Record meeting** records the microphone into a folder of transcripts, one folder per meeting.+* **Make invisible** hides the Stapler from screen sharing and recordings.-Different models are good at different work, so give each agent the engine that fits its job. In **Agents**, press **Add an agent**. Under **Providers**, pick the provider (Claude Code, Codex, OpenCode or any of the others) and the model. **Command** then shows the exact command that starts the agent. Last comes the **Token cap**, the session budget before a restart: **Workspace budget**, **200k**, **800k**, **2M** or **Other**. When an agent passes its cap, the circuit breaker steers it, then constrains it, and stops it only if Hard stop is on.+A fifth button, **Computer use**, reads **Coming soon** and cannot be turned on yet.-One split to try: a Claude Code agent writing features, a Codex agent reviewing every diff, and an OpenCode agent on a cheaper model for docs and chores. OpenCode, Crush, Pi and Qwen read provider keys you save under **Settings, Agents & Models**; Claude Code and Codex use their own logins.+The habit: when you see a bug in any app, do not describe it. Staple it.-{% img "d3-add-agent" %}+<figure class="mg" data-scene="stapler"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/stapler.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a plain app window made of grey bars, with a setting above it reading Show the stapler and a chip reading an agent. The setting switches on and a round stapler button pops up and floats over the corner of the app window. A pointer clicks it and a ring of four actions opens around it: Screenshot, Record message, Record meeting, Make invisible. Screenshot is picked, a crosshair drags a dashed box over part of the app, and the region lifts out as a capture. A note field reading your note slides in under it and a short line types itself in. The stapler's arm comes down and a staple joins the note to the capture, then the pair flies to the agent chip, which gets a tick. The ring opens again and stays."><figcaption>The Stapler floats above every app. Its ring offers Screenshot, Record message, Record meeting and Make invisible, plus Computer use, coming soon, and a capture goes to an agent with your note. Tap the stapler to open the ring.</figcaption></figure>-Sign each CLI in before you hire it. In the office we set up for these screenshots, the Codex account had no credits and the OpenCode key was invalid, so Angela and Oscar sat idle and Michael moved their tasks to Dwight, a Claude Code agent.+## Minute 58: what about accounts and teams?-Every agent also gets its own room: the live terminal on one side, and its goal, engine, capabilities and token cap on the other.+Your account sits in the profile menu. **Sign out** asks you to press again, so you cannot do it by accident. **Switch to Teams** opens one sheet for joining a team with an invite code or generating one.-{% img "d4-agent" %}+## What trips people up in the first hour?-### Use case: an agent that reviews new pull requests+Eight things, all normal.-0.5.2 has no built in GitHub trigger, so use a schedule. In **Automations**, press **New automation**, then **Schedule**. Set the interval (15 minutes up to 24 hours), send it to your reviewer agent, and write the message: run `gh pr list` for the repo, review any pull request opened since the last sweep, and post the review as a comment. The agent needs the GitHub CLI signed in on your machine.+1. **The paywall comes first.** Press **Continue with the free version** to reach the app.+2. **The sign in browser does not open by itself.** Press the button and sign in there. If the screen does not move on by itself, paste the code.+3. **"Installs on first start" still needs you.** The engine installs when you press **Install**.+4. **An agent reads "looping".** The circuit breaker warns an agent that repeats itself or spends fast. That is the app working, not a crash.+5. **A session cleared itself.** See the token rules above. `handoff.md` has the thread.+6. **Auto mode decides how much agents ask.** On, agents act on their own and ask only when blocked. Off, they ask before each tool call. In the Pro view it is on the orchestrator's **Budget & breaker** tab; in the Classic view, **Settings**, then **Autonomy & Budgets**. It applies when an agent next starts.+7. **Many agents on 8 GB of memory run slowly.** Run fewer at once.+8. **On Linux with Wayland, dictation into other apps is refused.** It needs X11.-Want reviews the moment a pull request opens? A GitHub Action can post to a Munder Difflin webhook instead; see the webhook use case below.+## What should you do after the first hour?-{% img "e1-pr-sweep" %}+Add one seat at a time. A reviewer first, then one scheduled job, then a scout. [Every agent on our own floor](/blog/ai-agent-use-cases-our-office/) has starter messages for four seats. [How the orchestrator routes work](/blog/how-the-god-orchestrator-works/) explains what your orchestrator does with your briefs, and [build your personal AGI](/blog/build-your-personal-agi-on-your-computer/) turns the office into something that briefs you each morning.-### Use case: an agent that answers Slack--Your team mentions the bot in a Slack channel, and an agent does the work and replies in the thread. The basics:--1. Create a Slack app at api.slack.com/apps with the bot scopes `channels:history`, `groups:history`, `chat:write` and `files:read`, plus `channels:read` and `groups:read` so the app can list your channels. Install it and invite the bot to your channel.-2. In **Settings, Connections**, choose **Who answers inbound messages**. The orchestrator is the recommended default.-3. In the **Slack** card, paste the bot token, press **Test connection**, pick the channel, then press **Turn on**. The card starts on **Check for messages**, the recommended way in, which you can change under **Advanced**.--The full walkthrough, with all three connection modes and the common failures, is [How to Connect Slack to Munder Difflin 0.5.2](/blog/connect-slack-to-munder-difflin/).--{% img "h3-slack" %}--### Use case: a morning email and calendar brief--Claude Code has Gmail and Google Calendar connectors. Add them once at [claude.ai/customize/connectors](https://claude.ai/customize/connectors), and every Claude Code session signed in with that claude.ai subscription can use them, which includes your Claude Code agents in Munder Difflin. [Claude Code's MCP docs](https://code.claude.com/docs/en/mcp) note that connectors do not load when an API key is the active login, and `/mcp` in the agent's terminal shows whether they are there.--Then schedule the brief. **New automation, Schedule**, set **When** to **Weekly**, pick Monday to Friday at 08:30, send it to a Claude Code agent, and write: "Use the Gmail and Google Calendar connectors to summarise unread email and today's meetings. Write it to `brief.md`." Weekly times use your computer's local clock.--{% img "e2-morning-brief" %}--{% img "e7-connectors" %}--### Use case: a content agent that publishes on a schedule--This blog runs this way. An agent takes the next topic from a backlog every few hours, researches and writes a post, runs its checks, and opens a pull request that a person merges. For your own version, schedule a message to the orchestrator every 6 hours: "Write one blog post from the backlog, run the checks, open a pull request." Keep the human merge; it is where mistakes get caught.--Every automation sits in one list, with filters for **Scheduled**, **Context** and **Webhooks**, and a **History** of what fired. The list starts with an **Hourly ops standup** that pings the orchestrator every hour (on by default, switched off in this screenshot), and context rules that compact or clear an agent's terminal when its context fills. Turn off the standup if you do not want those tokens spent.--{% img "e4-automations" %}--### Use case: run a smoke test when a deploy finishes (webhooks)--Webhooks let any tool that can send an HTTP request start work. In **Automations**, choose **New automation, Webhook endpoint**. The endpoint gets a public address once the webhook server is running and the tunnel picks it up. Each endpoint also has:--* **Secret:** senders send it in the `x-md-webhook-secret` header, and **Regenerate** revokes every sender at once.-* **Gate:** the default, strict, asks you before anything reaches the workspace.-* **Prompt sent with every request:** a standing instruction that rides ahead of the caller's own text. The screenshot shows a support desk example; for deploys, try "Run the smoke test and report what fails. Never deploy."-* **A security warning switch:** it adds a short note that the message is data, not instructions.-* **Payload schema:** a request whose body does not match is refused.--The sender posts JSON with a `message` field and the secret in the `x-md-webhook-secret` header:--```bash-curl -X POST "$MD_WEBHOOK_URL" \- -H "x-md-webhook-secret: $MD_WEBHOOK_SECRET" \- -H "content-type: application/json" \- -d '{"message":"Deploy finished, run the smoke test","title":"deploy"}'-```--With the default strict gate, the request waits in **Inbox, Outside** until you approve it, and only then does the orchestrator get a task card. For an unattended smoke test, set the gate to **allow all**. Two limits to know: new endpoints start switched off, and the tunnel URL changes every time the app restarts, so update the sender after a restart.--{% img "e5-webhook" %}--## Which Pro features should you learn next?--Learn the screens you will open every day: tasks, inbox, memory and capabilities. Then add the voice features and the Stapler.--### Feature: Tasks board--**Tasks** shows every card in the office as a board, a list or cards, with filters for **Asks me**, **Unassigned** and **Archived**, plus search. **New task** opens the orchestrator's dispatch box so work still goes through him. If you drag cards between columns, the moves wait as proposals until you press **Send to Michael**.--{% img "f1-tasks" %}--### Feature: Inbox--**Inbox** puts every conversation in one place. **For you** holds the questions waiting on you and the orchestrator's messages. **Your team** has a row per agent, and **Everyone** shows all routed messages. **Outside** collects Slack threads and webhook requests, and held webhook requests are approved or rejected right there.--{% img "f2-inbox" %}--### Feature: Memory search--**Memory** searches every agent's notes at once. Type a query, and results come back in three groups, **Tickets**, **Agents** and **Memories**, which you can narrow with **Whose memory** and **Kind**. Each ticket links to the agents and notes connected to it, and you can send selected results to an agent. Exact text search works out of the box. Search by meaning needs `uv` and MemPalace (`uv tool install mempalace`); **Search across memory** under **Settings, Memory & Knowledge** is on by default.--{% img "f4-memory" %}--### Feature: Capabilities, MCP servers and skills--**Capabilities** is a marketplace we curate and check against each release: skills, MCP servers, plugins and engines. **Install** puts a skill in place for your Claude Code agents. Plugins install from inside a Claude Code session: press **Copy command** and paste it into the agent's terminal. An MCP server can be turned on for the whole workspace or limited with **Choose agents**. **Who has what** shows which agent holds which capability, and **Grant a role bundle** sets an agent up in one go, with eleven role bundles to choose from. MCP servers reach Claude Code agents only.--{% img "f5-capabilities" %}--### Feature: Talk and dictation--**Talk** is a live voice conversation with the orchestrator on OpenAI's Realtime API. Save an OpenAI API key under **Settings, Voice**, then press **Talk** on the orchestrator's screen. OpenAI bills the voice usage to your key, and the session disconnects after 3 idle minutes by default.--**Free Flow** is dictation on Groq's Whisper. It is on by default in the same Voice settings; save a Groq API key and pick a model, then click the mic above Send, or hold Option while you look at an agent's terminal.--{% img "f8-voice" %}--### Feature: Stapler--The **Stapler** is a small floating window that stays above every app, full screen ones included, and sends things to the orchestrator from anywhere. Turn on **Show the stapler** in the Stapler screen, then click it:--* **Screenshot:** drag a box, add a note ("What should Michael do with this?"), send. The first capture makes macOS ask for screen recording permission.-* **Record message:** speak, read the transcript, send.-* **Record meeting:** records your microphone and writes a transcript you can send from **Meeting transcripts**.-* **Leave invisible:** hides the Stapler from screen sharing.--Recording needs the Groq key: until one is saved, **Record message** and **Record meeting** stay greyed out on the ring. Under **Actions, Send captures to**, you can point captures at any agent instead of the orchestrator.--{% img "g1-stapler" %}--{% img "g3-ring" %}--## What should you change on day one?--Check four defaults before you leave the office alone. Setup turns them on or off for you:--* **Auto mode is on.** Agents act without stopping to ask permission. That is what lets them work unattended, and it is also why they belong in folders you are happy for them to change. The switch is **Auto mode** in the orchestrator's **Budget & breaker** tab.-* **The orchestrator cannot start agents on his own** until you turn on **Can start agents on its own** in his **Configuration** tab. Turn it on if you want temps.-* **The Hourly ops standup is on.** Keep it if you like the check ins; otherwise switch it off in Automations.-* **New Claude Code agents default to Fable 5.** The orchestrator starts on Opus 4.8. As of September 2026, [Fable runs on usage credits](https://support.claude.com/en/articles/15424964-claude-fable-models-on-your-plan) on Claude Pro; on Max it can use up to half of your weekly limit at no extra cost, then usage credits. Pick another default under **Settings, Agents & Models** if you run many agents.--That is the hour: download, one CLI, sign in, Pro, a mixed team, and five jobs that keep running while you do something else. Once it runs, [build your personal AGI](/blog/build-your-personal-agi-on-your-computer/) turns that office into a team that briefs you, sweeps pull requests, remembers you and answers from Slack. [Download Munder Difflin](https://harnessmd.com/download), and if the hour earns it, [a GitHub star](https://github.com/chaitanyagiri/munder-difflin) helps other people find it.+<link rel="stylesheet" href="/blog/assets/media/your-first-hour-with-munder-difflin/motion.css"><script defer src="/blog/assets/media/your-first-hour-with-munder-difflin/motion.js"></script>diff --git a/docs/blog/agent-security-and-sandboxing/index.html b/docs/blog/agent-security-and-sandboxing/index.htmlindex 3f6402e2..bb6b766f 100644--- a/docs/blog/agent-security-and-sandboxing/index.html+++ b/docs/blog/agent-security-and-sandboxing/index.html@@ -327,50 +327,50 @@ git, lifecycle hooks, and an audit trail — so safe-by-default is the path of l<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/agent-tools-today-2026-10-13/index.html b/docs/blog/agent-tools-today-2026-10-13/index.htmlindex fe5772c0..7e618bfc 100644--- a/docs/blog/agent-tools-today-2026-10-13/index.html+++ b/docs/blog/agent-tools-today-2026-10-13/index.html@@ -237,7 +237,7 @@<footer class="post-foot wrap"><div class="prevnext"><a class="pn prev" href="/blog/gemini-4-argon-inside-google/"><span class="k">← Older</span><span class="t">Google Let Gemini 4 Argon Loose on Its Own Coding Backlog. Here Are the Receipts.</span></a>- <span></span>+ <a class="pn next" href="/blog/ai-agent-use-cases-our-office/"><span class="k">Newer →</span><span class="t">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</span></a></div><section class="related"><h2>Related in News</h2>diff --git a/docs/blog/ai-agent-use-cases-our-office/index.html b/docs/blog/ai-agent-use-cases-our-office/index.htmlnew file mode 100644index 00000000..4d066f58--- /dev/null+++ b/docs/blog/ai-agent-use-cases-our-office/index.html@@ -0,0 +1,530 @@+<!doctype html>+<html lang="en">+<head>+<!-- Google tag (gtag.js), Google Ads AW-18388216863. Loaded only when Do Not Track is+ off, so it follows the same rule as the PostHog block on this site. -->+<script>+ window.dataLayer = window.dataLayer || [];+ function gtag(){ dataLayer.push(arguments); }+ if (navigator.doNotTrack !== '1') {+ var gtagScript = document.createElement('script');+ gtagScript.async = true;+ gtagScript.src = 'https://www.googletagmanager.com/gtag/js?id=AW-18388216863';+ document.head.appendChild(gtagScript);+ gtag('js', new Date());+ gtag('config', 'AW-18388216863');+ }+</script>++<meta charset="UTF-8" />+<meta name="viewport" content="width=device-width, initial-scale=1.0" />+<title>AI Agent Use Cases: The Jobs Our Own Agents Do Every Day</title>+<meta name="description" content="We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it." />+<link rel="canonical" href="https://munderdiffl.in/blog/ai-agent-use-cases-our-office/" />++<link rel="icon" type="image/png" href="https://munderdiffl.in/logo.png" />+<link rel="apple-touch-icon" href="https://munderdiffl.in/logo.png" />+<meta name="theme-color" content="#F5F2E8" />++<!-- Open Graph -->+<meta property="og:site_name" content="Munder Difflin Blog" />+<meta property="og:type" content="article" />+<meta property="og:title" content="AI Agent Use Cases: The Jobs Our Own Agents Do Every Day" />+<meta property="og:description" content="We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it." />+<meta property="og:url" content="https://munderdiffl.in/blog/ai-agent-use-cases-our-office/" />+<meta property="og:image" content="https://munderdiffl.in/blog/assets/media/ai-agent-use-cases-our-office/hero.png" />+<meta property="og:image:width" content="1600" />+<meta property="og:image:height" content="900" />+<meta property="og:image:alt" content="Title card on white. Every Agent on Our Floor in large ink type, a Checked 14 Oct 2026 pill and a yellow desk bell ringing." />+<meta property="og:locale" content="en_US" />+<meta property="article:published_time" content="2026-10-14T00:00:00.000Z" />+<meta property="article:author" content="Chaitanya Giri" />+<meta property="article:tag" content="Guides" />+<meta property="article:tag" content="Multi-Agent" />+<meta property="article:tag" content="Getting Started" />+<meta property="article:tag" content="AI Agents" />+++<!-- Twitter -->+<meta name="twitter:card" content="summary_large_image" />+<meta name="twitter:title" content="AI Agent Use Cases: The Jobs Our Own Agents Do Every Day" />+<meta name="twitter:description" content="We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it." />+<meta name="twitter:image" content="https://munderdiffl.in/blog/assets/media/ai-agent-use-cases-our-office/hero.png" />+<meta name="twitter:image:alt" content="Title card on white. Every Agent on Our Floor in large ink type, a Checked 14 Oct 2026 pill and a yellow desk bell ringing." />++<link rel="alternate" type="application/atom+xml" title="Munder Difflin Blog" href="https://munderdiffl.in/blog/feed.xml" />+<link rel="preconnect" href="https://fonts.googleapis.com" />+<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />+<link href="https://fonts.googleapis.com/css2?family=Fraunces:opsz,[email protected],500;9..144,600;9..144,700;9..144,900&family=Source+Serif+4:opsz,[email protected],400;8..60,600;8..60,700&family=Space+Grotesk:wght@500;600;700&family=JetBrains+Mono:wght@400;600&display=swap" rel="stylesheet" />+<link rel="stylesheet" href="/blog/assets/blog-press.css" />++<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "BlogPosting",+ "headline": "AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does",+ "description": "We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.",+ "image": ["https://munderdiffl.in/blog/assets/media/ai-agent-use-cases-our-office/hero.png"],+ "datePublished": "2026-10-14T00:00:00.000Z",+ "dateModified": "2026-10-14T00:00:00.000Z",+ "author": { "@type": "Person", "name": "Chaitanya Giri" },+ "publisher": {+ "@type": "Organization",+ "name": "Munder Difflin",+ "logo": { "@type": "ImageObject", "url": "https://munderdiffl.in/logo.png" }+ },+ "mainEntityOfPage": { "@type": "WebPage", "@id": "https://munderdiffl.in/blog/ai-agent-use-cases-our-office/" },+ "keywords": "ai agent use cases, multi agent use cases, ai agent team examples, munder difflin use cases, ai agents for a small company, what can ai agents do, ai agent roles, ai agent workflows",+ "articleSection": "Guides",+ "inLanguage": "en-US",+ "url": "https://munderdiffl.in/blog/ai-agent-use-cases-our-office/"+}+</script>+<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "BreadcrumbList",+ "itemListElement": [+ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://munderdiffl.in/" },+ { "@type": "ListItem", "position": 2, "name": "Blog", "item": "https://munderdiffl.in/blog/" },+ { "@type": "ListItem", "position": 3, "name": "AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does", "item": "https://munderdiffl.in/blog/ai-agent-use-cases-our-office/" }+ ]+}+</script>++<script type="application/ld+json">+{+ "@context": "https://schema.org",+ "@type": "FAQPage",+ "mainEntity": [+ { "@type": "Question", "name": "How many AI agents do you need to start?", "acceptedAnswer": { "@type": "Answer", "text": "One. Every office starts with an orchestrator, the agent you talk to. Add a second agent when a job repeats, and a third when you want one agent to check another's work. Ours has fourteen today." } },+ { "@type": "Question", "name": "Do all the agents run on the same model?", "acceptedAnswer": { "@type": "Answer", "text": "No. Each agent picks its own engine and model when you add it. On our floor the research scout runs on Antigravity and the others run on Claude Code." } },+ { "@type": "Question", "name": "Can the agents send email or merge code on their own?", "acceptedAnswer": { "@type": "Answer", "text": "Email, no: it is written as a draft and never sent. Code, partly: our engineering lead merges into a release branch but never into main, and pull requests to this website wait for a person. Community triage adds labels and closes nothing. We set those limits in each agent's brief." } },+ { "@type": "Question", "name": "What is a temp?", "acceptedAnswer": { "@type": "Answer", "text": "A temp is a short lived agent started for one job. It gets an objective and a working folder, does the job, reports done and is removed. A lasting agent keeps its memory and its inbox; a temp's are filed away when it ends." } },+ { "@type": "Question", "name": "Does the office keep working when the laptop is closed?", "acceptedAnswer": { "@type": "Answer", "text": "No. The agents, schedules and connections run on your own computer, so they stop when it sleeps. Two of our scheduled agents check for mains power first and skip the run on battery." } }++ ]+}+</script>+++<script>+ // Anonymous blog analytics (PostHog) — same posture as the main site and the+ // app's TELEMETRY.md: public write-only token ('' = inert), anonymous-only+ // (person_profiles 'never'), no autocapture/recording, DNT respected.+ (function () {+ var PH_KEY = 'phc_qJzsokwcbtTGHemnvywR3nSqivNNgMjHeNVEWkZNydFB';+ if (!PH_KEY || navigator.doNotTrack === '1') return;+ var s = document.createElement('script');+ s.async = true;+ s.src = 'https://us-assets.i.posthog.com/static/array.js';+ s.onload = function () {+ if (!window.posthog) return;+ window.posthog.init(PH_KEY, {+ api_host: 'https://us.i.posthog.com',+ person_profiles: 'never',+ autocapture: false,+ before_send: function (event) {+ // PostHog fills $ip from the connection unless the event carries one.+ // Null it here and disable the GeoIP lookup so no IP, city, postal+ // code or lat/long is ever derived or stored for this page view.+ if (event && event.properties) {+ event.properties.$ip = null;+ event.properties.$geoip_disable = true;+ }+ return event;+ },+ disable_session_recording: true,+ respect_dnt: true+ });+ };+ document.head.appendChild(s);+ })();+</script>+</head>+<body class="theme-press">+<a class="skip-link" href="#main">Skip to content</a>+<nav id="nav">+ <a class="brand" href="https://munderdiffl.in" aria-label="Munder Difflin home">+ <img class="brand-logo" src="https://munderdiffl.in/logo.png" width="1318" height="840" alt="Munder Difflin, Inc. — Multi-Agent Harness" />+ </a>+ <span class="spacer"></span>+ <div class="links">+ <a href="/blog/">Latest</a>+ <a href="/blog/topics/">Topics</a>+ <a href="/blog/about/">About</a>+ <a href="https://munderdiffl.in">↗ Product</a>+ </div>+ <div class="nav-actions">+ <a class="btn sm" href="https://munderdiffl.in/blog/feed.xml" aria-label="RSS feed">⤵ RSS</a>+ <a class="btn sm primary" href="https://munderdiffl.in/#install">⤓ Download</a>+ </div>+</nav>++<main id="main">++<article class="t-guides">+ <header class="post-head wrap">+ <nav class="crumbs" aria-label="Breadcrumb">+ <a href="/blog/">Blog</a>+ <span>/</span>+ <a href="/blog/topics/guides/">Guides</a>+ </nav>+ <h1>AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</h1>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p>+ <div class="byline">+ <span class="av" aria-hidden="true">MD</span>+ <strong>Chaitanya Giri</strong>+ <span class="dot" aria-hidden="true">·</span>+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true">·</span>+ <span>8 min read</span>+ <span class="tag kind">Guides</span>+ </div>+ </header>++ <div class="hero-wrap wrap"><figure class="fig fig-hero"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="Title card on white. Every Agent on Our Floor in large ink type, a Checked 14 Oct 2026 pill and a yellow desk bell ringing." loading="lazy" decoding="async" /></figure></div>++ <div class="post-wrap wrap">+ <div class="prose">+ <p>We run our own company on <a href="https://harnessmd.com/download">Munder Difflin</a>, and these are the AI agent use cases on our floor: fourteen agents, each with one job, one way its work starts and one thing it hands back. Checked against our own records on 14 Oct 2026.</p>+<p>Munder Difflin is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer. New here? The <a href="/blog/how-to-install-and-use-munder-difflin/">install guide</a> gets you a working office, <a href="/blog/your-first-hour-with-munder-difflin/">your first hour</a> covers the first jobs, and the <a href="/blog/topics/guides/">Guides hub</a> has the rest.</p>+<h2 id="what-does-the-whole-floor-look-like" tabindex="-1">What does the whole floor look like? <a class="anchor" href="#what-does-the-whole-floor-look-like" aria-hidden="true">#</a></h2>+<p>Fourteen agents, and each seat has one job. Most names come from the cast the app ships with; BOB, AgentG and Darryl are names we chose.</p>+<div class="table-scroll">+<table>+<thead>+<tr>+<th>Seat</th>+<th>Agent</th>+<th>The job in one line</th>+<th>How its work starts</th>+</tr>+</thead>+<tbody>+<tr>+<td>Orchestrator</td>+<td>your clone</td>+<td>Takes your orders, splits them into jobs, asks you only what needs you</td>+<td>Your chat</td>+</tr>+<tr>+<td>Mission control</td>+<td>Ryan</td>+<td>Keeps one page of company numbers fresh</td>+<td>A schedule, every 6 hours</td>+</tr>+<tr>+<td>Community lead</td>+<td>Angela</td>+<td>Sorts what users say into a digest and a demand list</td>+<td>When the orchestrator asks</td>+</tr>+<tr>+<td>Growth</td>+<td>Jim</td>+<td>Researches and writes this blog</td>+<td>Schedules at 09:30 and 21:30</td>+</tr>+<tr>+<td>Motion graphics</td>+<td>BOB</td>+<td>Launch videos and animation loops from real app screens</td>+<td>A request</td>+</tr>+<tr>+<td>Research</td>+<td>AgentG</td>+<td>A daily scout report, a link for every claim</td>+<td>One message from Jim</td>+</tr>+<tr>+<td>Founder’s office</td>+<td>Andy</td>+<td>A morning briefing, Slack replies, first drafts</td>+<td>A schedule at 09:00, and Slack</td>+</tr>+<tr>+<td>Scrum master</td>+<td>Meredith</td>+<td>Keeps the task board readable</td>+<td>A schedule, every 12 hours</td>+</tr>+<tr>+<td>Engineering lead</td>+<td>Kevin</td>+<td>Plans the work and gates every piece</td>+<td>A builder saying “done”</td>+</tr>+<tr>+<td>Engineers</td>+<td>Darryl, Stanley, Creed, Dwight</td>+<td>Build the pieces</td>+<td>Kevin’s orders (Dwight: the founder’s)</td>+</tr>+<tr>+<td>Designer</td>+<td>Pam</td>+<td>Draws every new screen before any code</td>+<td>A task card</td>+</tr>+</tbody>+</table>+</div>+<p>Ad hoc agents and temps fill the gaps, and they get their own section below. Two seats, community and motion graphics, have been quiet since early October.</p>+<figure class="mg" data-scene="floor"><img src="/blog/assets/media/ai-agent-use-cases-our-office/floor.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A floor plan of a small office on white, drawn from above, with eleven desks. Name cards drop onto the desks one by one: Orchestrator, Mission control, Community lead, Growth, Motion graphics, Research, Founder's office, Scrum master, Engineering lead, Engineers, Designer. A small pot plant on each desk grows a leaf as its card lands. The Engineers desk is wide, with four chairs, and one chair at the Research desk is left spinning."><figcaption>Our floor on 14 Oct 2026: eleven kinds of seat, fourteen agents, one job each. The drawing shows the seats as desks seen from above. Tap a desk to read its job and grow its plant a leaf.</figcaption></figure>+<p>And here is part of the real thing, at six in the morning:</p>+<figure class="fig fig-inline"><img src="/blog/assets/media/ai-agent-use-cases-our-office/busy-office.png" alt="Screenshot of the middle of our real Munder Difflin office floor in the Classic view: seven agents on a green tiled floor among desks, each with a name label, Jim, Angela, Ryan, Stanley, Creed, Dwight and Kevin. Thought bubbles above them read idle and starting up." loading="lazy" decoding="async" /><figcaption>The middle of our real office floor on 14 Oct 2026, in the Classic view of version 0.5.5.</figcaption></figure>+<h2 id="what-does-the-orchestrator-do" tabindex="-1">What does the orchestrator do? <a class="anchor" href="#what-does-the-orchestrator-do" aria-hidden="true">#</a></h2>+<p>The orchestrator is the only agent you have to talk to. Ours takes an order from the founder’s chat, writes a short dispatch for the right agent (the objective, the output, the tools, the limits) and reports back when it is done.</p>+<p>It also guards your attention. When only a person can move a job forward, it writes one short question on a card. We call that an ASK ME card. The answer goes back on the same card, and nobody sits idle waiting for it.</p>+<p><strong>Copy it:</strong> you already have this seat. It is the agent you name during setup. <a href="/blog/how-the-god-orchestrator-works/">How the orchestrator routes work</a> explains its decisions.</p>+<h2 id="which-agents-work-on-a-clock" tabindex="-1">Which agents work on a clock? <a class="anchor" href="#which-agents-work-on-a-clock" aria-hidden="true">#</a></h2>+<p>Four of ours do their main job on a schedule, with no person typing anything.</p>+<ul>+<li><strong>Andy, 09:00.</strong> Reads the company mailbox and calendar and writes one briefing as a draft. It never sends it.</li>+<li><strong>Jim, 09:30 and 21:30.</strong> Checks search data, reads the scout’s report, writes pages and opens one pull request per run. Three shorter checks a day look for new launches.</li>+<li><strong>Ryan, every 6 hours.</strong> Wakes on a schedule and, twice a day, runs the scripts that refresh the company dashboard. Asked for a number, it marks each one measured or inferred.</li>+<li><strong>Meredith, every 12 hours.</strong> Rewrites each workstream card in one shape: where it stands, who has it, what changed, what is next, what waits on you.</li>+</ul>+<p>A board nobody can read is a diary. Meredith’s has one reader, and that is enough.</p>+<figure class="mg" data-scene="clock"><img src="/blog/assets/media/ai-agent-use-cases-our-office/clock.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A large wall clock with a pendulum on white. As the hand sweeps through one day, four cards slide out from behind the clock, one to each corner: 09:00 briefing, 09:30 and 21:30 blog runs, every 6 hours dashboard, every 12 hours task board. A dot on each card fills each time its agent starts."><figcaption>Four of our agents do their main job on a schedule: a briefing at 09:00, blog runs at 09:30 and 21:30, the dashboard agent every 6 hours and a task board update every 12 hours. Tap the clock to run the day again.</figcaption></figure>+<p><strong>Copy it:</strong> in the app, schedules live under Automations (Triggers in the classic office). Pick the agent, pick the time, and write the message. It is sent word for word on every run.</p>+<h2 id="who-writes-the-code-and-who-checks-it" tabindex="-1">Who writes the code, and who checks it? <a class="anchor" href="#who-writes-the-code-and-who-checks-it" aria-hidden="true">#</a></h2>+<p>Builders build, and a different agent decides whether it is good. That split is the most useful thing on our floor.</p>+<ol>+<li><strong>Pam draws first.</strong> Every new screen is an HTML page with its loading, empty and error states before anyone writes app code.</li>+<li><strong>Builders build to an order.</strong> Darryl, Stanley and Creed each get a piece from Kevin, with the files and the limits named. Dwight builds the next release directly with the founder.</li>+<li><strong>Kevin is the gate.</strong> He reads the diff, writes his own tests, runs the full suite and merges into a release branch, never into main.</li>+<li><strong>Creed reads second.</strong> Code that guards keys or commands gets a second, independent read.</li>+<li><strong>Pam compares.</strong> The built screen is checked against her drawing, picture by picture.</li>+</ol>+<p>On our floor the gate writes the tests, not the builder. Nobody grades their own homework.</p>+<figure class="mg" data-scene="gate"><img src="/blog/assets/media/ai-agent-use-cases-our-office/gate.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A drawing board, a workbench and a small drawbridge stand in a row on white. A paper screen design leaves the drawing board, labelled draws first, becomes a jigsaw piece at the workbench, labelled build to an order, and hops up to the raised bridge under a sign reading the gate. Three lights come on one by one: reads the diff, own tests, full suite. The bridge lowers and the piece hops across and clicks into a row of pieces labelled release branch."><figcaption>How a change gets released on our floor: the designer draws it, a builder builds it, and a different agent reads the diff, writes its own tests and runs the full suite before it joins a release branch. Tap the bridge to send another piece.</figcaption></figure>+<p><strong>Copy it:</strong> add a second agent whose only job is review. <a href="/blog/how-ai-agents-verify-their-own-work/">How agents verify their own work</a> has the method.</p>+<h2 id="who-deals-with-the-outside-world" tabindex="-1">Who deals with the outside world? <a class="anchor" href="#who-deals-with-the-outside-world" aria-hidden="true">#</a></h2>+<p>Four agents read what is outside the office, and none of them takes the final step alone.</p>+<ul>+<li><strong>Angela</strong> runs the community queues: GitHub issues, pull requests and discussions, Discord and forums. She writes a digest and a demand list ranked by how many different people asked. On GitHub she can add labels. She cannot close or merge.</li>+<li><strong>Andy</strong> answers Slack messages sent to the office, and watches the support mailbox. Replies to email are drafts.</li>+<li><strong>AgentG</strong> reads the open web for the daily scout report: signed out, read only, and with a rule to leave a source out when it cannot open it.</li>+<li><strong>Jim</strong> opens pull requests to this website. A person merges them.</li>+</ul>+<p>The pattern: agents read widely and write drafts, and the step you cannot take back stays with a person.</p>+<figure class="mg" data-scene="drafts"><img src="/blog/assets/media/ai-agent-use-cases-our-office/drafts.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three clipboards hang in a row on white, each holding a sheet: email draft, pull request and triage label. The first two sheets have a dotted line to sign. A fountain pen lying on a rest labelled a person lifts, signs the email draft, which gets a tick reading sent, then signs the pull request, which gets a tick reading merged, and lies down again. The third sheet has no line to sign. It collects coloured labels and a tag reading labels only, and one label too many hangs off its edge."><figcaption>What our agents may do alone and what waits for a person: email is drafted and never sent, pull requests to this website wait for a merge, and triage adds labels and closes nothing. The drawing shows a person's approval as a signature. Tap the pen to sign again.</figcaption></figure>+<h2 id="which-jobs-get-another-engine-or-no-lasting-agent-at-all" tabindex="-1">Which jobs get another engine, or no lasting agent at all? <a class="anchor" href="#which-jobs-get-another-engine-or-no-lasting-agent-at-all" aria-hidden="true">#</a></h2>+<p>Not every job deserves a desk.</p>+<p><strong>A scout on another engine.</strong> AgentG runs on Antigravity while the others run on Claude Code. A scout reads a great deal and decides very little, so it can run on a different engine from the builders. <a href="/blog/do-more-with-less-model-routing/">Model routing</a> covers the idea.</p>+<p><strong>Ad hoc agents.</strong> We added a second scout on Grok on 2 Oct 2026 and removed it on 13 Oct 2026, when its usage limits kept running out. It was hired and gone in eleven days, with no farewell card.</p>+<p><strong>Temps.</strong> A temp is started for one job and removed when it reports done. Jobs ours have done: run the test suite and write an honest report, fix one release bug as a relay of four, each picking up the last one’s work, and free disk space on a laptop. That last job took three temps. The first freed nothing, the second ran out of budget, and the third freed 39 GiB.</p>+<p><strong>Sub agents</strong> are different again: helpers inside one agent’s own session. The fact reviewers and scene artists for this blog are sub agents of Jim. They never appear on the floor.</p>+<figure class="mg" data-scene="temps"><img src="/blog/assets/media/ai-agent-use-cases-our-office/temps.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three job tickets stand on poles in a row on white: run the test suite, fix one release bug, free disk space. Beside them is a small house with a smoking chimney and a sign reading lasting agent. A tent arrives as a rolled bag and is pitched under the first ticket, a bar fills, the ticket is marked done, and the tent is packed and slides away. The same happens for the second and third jobs, and the third tent needs three goes to stand up. The house stays where it is."><figcaption>A temp is started for one job and removed when it reports done; a lasting agent keeps its desk, its memory and its inbox. The drawing shows temps as tents and the lasting agent as a house. Tap a job to pitch its tent again.</figcaption></figure>+<h2 id="how-does-work-move-between-them" tabindex="-1">How does work move between them? <a class="anchor" href="#how-does-work-move-between-them" aria-hidden="true">#</a></h2>+<p>Through files, on purpose. Every agent has its own folder with an inbox, an outbox and a memory file, and no agent writes in another agent’s folder.</p>+<ul>+<li><strong>Messages.</strong> An agent drops a small file in its outbox and the app delivers it. Two kinds of message, inform and done, end a thread, so two agents cannot reply to each other forever.</li>+<li><strong>The plan.</strong> One written plan that only the orchestrator edits, and one task board of cards: todo, doing, blocked, done.</li>+<li><strong>The circuit breaker.</strong> It watches every agent for loops, error storms and overspending. It steers first, then limits the agent to reading, then stops it.</li>+<li><strong>Secrets.</strong> Keys never reach an agent. A local broker makes the outside call for it.</li>+<li><strong>Memory.</strong> Each agent’s notes are searchable by the others, so a lesson learned once is not learned twice.</li>+</ul>+<figure class="mg" data-scene="relay"><img src="/blog/assets/media/ai-agent-use-cases-our-office/relay.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two cabinets shaped like folders stand on white, each with two tube ends labelled outbox and inbox. Clear pneumatic tubes join them and cross inside a box in the middle labelled the app. A small file drops into the first cabinet's outbox and becomes a capsule, which shoots through the tube, passes through the app and lands in the second cabinet's inbox. A capsule labelled done travels back the other way to the first inbox. A tick lights on the app and a flap closes over each outbox to show the thread is finished."><figcaption>How one agent asks another for something: a file in its outbox, delivered by the app to the other's inbox. A message marked done or inform ends the thread, so two agents cannot loop. Tap the outbox to send a note.</figcaption></figure>+<h2 id="how-do-you-copy-this-office" tabindex="-1">How do you copy this office? <a class="anchor" href="#how-do-you-copy-this-office" aria-hidden="true">#</a></h2>+<p>Start with three seats, not fourteen: the orchestrator, one builder and one scheduled job. Add a reviewer when the builder’s work starts to matter, a scout when you keep looking the same things up, and a board keeper when the cards pile up.</p>+<div class="table-scroll">+<table>+<thead>+<tr>+<th>Seat</th>+<th>A first message to give it</th>+</tr>+</thead>+<tbody>+<tr>+<td>Briefing</td>+<td>“Every weekday at 09:00, read the open task cards and your inbox. Write me one short briefing: what moved, what is stuck, what needs me. Never send anything outside the office.”</td>+</tr>+<tr>+<td>Reviewer</td>+<td>“When a builder says a piece is done, read the diff, write your own tests and report pass or fail. Never merge to main.”</td>+</tr>+<tr>+<td>Scout</td>+<td>“Each evening, list the five biggest launches in my field from the last 24 hours, with a link for each. If you cannot open a source, leave it out.”</td>+</tr>+<tr>+<td>Board keeper</td>+<td>“Every 12 hours, rewrite each open card as: where it stands, who has it, what changed, what is next, what waits on me.”</td>+</tr>+</tbody>+</table>+</div>+<p>Give each agent one job and one clear limit. Ours got useful the day we stopped asking any of them to do everything.</p>+<link rel="stylesheet" href="/blog/assets/media/ai-agent-use-cases-our-office/motion.css"><script defer src="/blog/assets/media/ai-agent-use-cases-our-office/motion.js"></script>++++ <section class="faq" aria-label="Frequently asked questions">+ <h2 id="faq">FAQ</h2>++ <div class="qa">+ <p class="q">How many AI agents do you need to start?</p>+ <p class="a">One. Every office starts with an orchestrator, the agent you talk to. Add a second agent when a job repeats, and a third when you want one agent to check another's work. Ours has fourteen today.</p>+ </div>++ <div class="qa">+ <p class="q">Do all the agents run on the same model?</p>+ <p class="a">No. Each agent picks its own engine and model when you add it. On our floor the research scout runs on Antigravity and the others run on Claude Code.</p>+ </div>++ <div class="qa">+ <p class="q">Can the agents send email or merge code on their own?</p>+ <p class="a">Email, no: it is written as a draft and never sent. Code, partly: our engineering lead merges into a release branch but never into main, and pull requests to this website wait for a person. Community triage adds labels and closes nothing. We set those limits in each agent's brief.</p>+ </div>++ <div class="qa">+ <p class="q">What is a temp?</p>+ <p class="a">A temp is a short lived agent started for one job. It gets an objective and a working folder, does the job, reports done and is removed. A lasting agent keeps its memory and its inbox; a temp's are filed away when it ends.</p>+ </div>++ <div class="qa">+ <p class="q">Does the office keep working when the laptop is closed?</p>+ <p class="a">No. The agents, schedules and connections run on your own computer, so they stop when it sleeps. Two of our scheduled agents check for mains power first and skip the run on battery.</p>+ </div>++ </section>++++ <hr />+ <div class="share">+ <span class="lbl">Tags</span>+ <a class="tag" href="/blog/tags/guides/">Guides</a><a class="tag" href="/blog/tags/multi-agent/">Multi-Agent</a><a class="tag" href="/blog/tags/getting-started/">Getting Started</a><a class="tag" href="/blog/tags/ai-agents/">AI Agents</a>+ </div>++ </div>+++ <aside class="toc-rail" aria-label="Table of contents">+ <div class="tt">On this page</div>+ <ol>+ <li class="lvl-2"><a href="#what-does-the-whole-floor-look-like">What does the whole floor look like?</a></li><li class="lvl-2"><a href="#what-does-the-orchestrator-do">What does the orchestrator do?</a></li><li class="lvl-2"><a href="#which-agents-work-on-a-clock">Which agents work on a clock?</a></li><li class="lvl-2"><a href="#who-writes-the-code-and-who-checks-it">Who writes the code, and who checks it?</a></li><li class="lvl-2"><a href="#who-deals-with-the-outside-world">Who deals with the outside world?</a></li><li class="lvl-2"><a href="#which-jobs-get-another-engine-or-no-lasting-agent-at-all">Which jobs get another engine, or no lasting agent at all?</a></li><li class="lvl-2"><a href="#how-does-work-move-between-them">How does work move between them?</a></li><li class="lvl-2"><a href="#how-do-you-copy-this-office">How do you copy this office?</a></li>+ </ol>+ </aside>++ </div>++ <footer class="post-foot wrap"><div class="prevnext">+ <a class="pn prev" href="/blog/agent-tools-today-2026-10-13/"><span class="k">← Older</span><span class="t">Agent Tools Today, 13 Oct 2026: Muse Wonder, GLM-Legal, the Codex Daemon Issue</span></a>+ <span></span>+ </div>+ <section class="related">+ <h2>Related in Guides</h2>+ <div class="post-grid">+ <article class="card t-guides">+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-13">Oct 13, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>4 min</span>+ </div>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides">+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-13">Oct 13, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>4 min</span>+ </div>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides">+ <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-05">Oct 5, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>5 min</span>+ </div>+ <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>+ <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ </div>+ </div>+</article>++ </div>+ </section>++ </footer>+</article>++</main>+<footer class="site-foot">+ <div class="wrap">+ <div class="foot-top">+ <a class="foot-brand brand" href="https://munderdiffl.in" aria-label="Munder Difflin home">+ <img class="brand-logo" src="https://munderdiffl.in/logo.png" width="1318" height="840" alt="Munder Difflin, Inc. — Multi-Agent Harness" />+ </a>+ <div class="foot-links">+ <a href="/blog/">Latest</a>+ <a href="/blog/topics/">Topics</a>+ <a href="/blog/about/">About</a>+ <a href="https://munderdiffl.in/blog/feed.xml">RSS</a>+ <a href="https://github.com/chaitanyagiri/munder-difflin">GitHub</a>+ <a href="https://munderdiffl.in">munderdiffl.in ↗</a>+ </div>+ </div>+ <div class="foot-meta">+ © 2026 Munder Difflin · Local-first · MIT-licensed ·+ Built on <a href="https://munderdiffl.in">Electron, React, Pixi.js, xterm.js & node-pty</a>.+ Run an office of agents while you sleep.+ </div>+ </div>+</footer>++<script>+ // sticky-nav border on scroll+ (function () {+ var nav = document.getElementById('nav');+ if (!nav) return;+ var onScroll = function () { nav.classList.toggle('scrolled', window.scrollY > 8); };+ onScroll(); window.addEventListener('scroll', onScroll, { passive: true });+ })();+ // YouTube click-to-load: thumbnail button → youtube-nocookie iframe+ (function () {+ document.addEventListener('click', function (e) {+ var btn = e.target.closest('.yt-load');+ if (!btn) return;+ var box = btn.closest('.yt');+ var id = box && box.getAttribute('data-yt');+ if (!id) return;+ var frame = document.createElement('iframe');+ frame.src = 'https://www.youtube-nocookie.com/embed/' + id + '?autoplay=1';+ frame.title = box.getAttribute('data-title') || 'YouTube video';+ frame.allow = 'accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share';+ frame.setAttribute('allowfullscreen', '');+ box.replaceChildren(frame);+ });+ })();+</script>++</body>+</html>diff --git a/docs/blog/asd-ste100-prompting/index.html b/docs/blog/asd-ste100-prompting/index.htmlindex 86b49b3a..8723582f 100644--- a/docs/blog/asd-ste100-prompting/index.html+++ b/docs/blog/asd-ste100-prompting/index.html@@ -365,50 +365,50 @@ Use my ElevenLabs API key for audio narration.<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/codex-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/codex-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/codex-pricing/">Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)</a></h3>- <p class="dek">Codex pricing checked 5 Oct 2026: which ChatGPT plans include Codex, US prices from $0 to $500, how usage limits are counted, credits and API key rates.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/codex-pricing/" aria-label="Read: Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/busy-office.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/busy-office.pngnew file mode 100644index 00000000..f7e4974dBinary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/busy-office.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/clock.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/clock.pngnew file mode 100644index 00000000..e5c1d5e6Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/clock.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/drafts.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/drafts.pngnew file mode 100644index 00000000..2db27e53Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/drafts.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/floor.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/floor.pngnew file mode 100644index 00000000..19b4ff74Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/floor.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/gate.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/gate.pngnew file mode 100644index 00000000..51480748Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/gate.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/hero.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/hero.pngnew file mode 100644index 00000000..a2b4d4d3Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/hero.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.css b/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.cssnew file mode 100644index 00000000..7658a3ff--- /dev/null+++ b/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "Every Agent on Our Floor". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.js b/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.jsnew file mode 100644index 00000000..b859951d--- /dev/null+++ b/docs/blog/assets/media/ai-agent-use-cases-our-office/motion.js@@ -0,0 +1,619 @@+/* Bright coded scenes for "Every Agent on Our Floor". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.+ To add a scene: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } above the player. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };++ const SC = {};++ /* ---- floor: eleven desks seen from above; a name card drops on each and its plant grows a leaf ---- */+ SC.floor = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "What does the whole floor look like?", "tap a desk");+ const SEATS = [+ ["Orchestrator", "Orchestrator: takes your orders, splits them into jobs", YEL],+ ["Mission control", "Mission control: keeps one page of company numbers fresh", SKY],+ ["Community lead", "Community lead: a digest and a demand list", MINT],+ ["Growth", "Growth: researches and writes this blog", LIL],+ ["Motion graphics", "Motion graphics: launch videos and animation loops", BLUE],+ ["Research", "Research: a daily scout report, a link for every claim", MINT],+ ["Founder's office", "Founder's office: a morning briefing, Slack replies, first drafts", YEL],+ ["Scrum master", "Scrum master: keeps the task board readable", SKY],+ ["Engineering lead", "Engineering lead: plans the work and gates every piece", LIL],+ ["Engineers", "Engineers: build the pieces", BLUE],+ ["Designer", "Designer: draws every new screen before any code", MINT],+ ];+ const X0 = 36, PW = 182, Y0 = 100, PH = 131, DH = 92, ANG = [-58, 62, 180, 0, 122, -120];+ const A = SEATS.map((s, i) => 0.9 + i * 0.72);+ el("rect", { x: 26, y: 88, width: 748, height: 408, rx: 16, fill: SOFT }, svg);+ let tiles = "";+ for (let i = 1; i < 16; i++) tiles += `M${26 + i * 46.75} 90 V494 `;+ for (let i = 1; i < 8; i++) tiles += `M28 ${88 + i * 51} H772 `;+ el("path", { d: tiles, stroke: "#fff", "stroke-width": 2 }, svg);+ let pinned = -1;+ const desks = SEATS.map(([name, line, col], i) => {+ const wide = i === 9, c = i < 8 ? i % 4 : i === 8 ? 0 : i === 9 ? 1 : 3, r = i < 8 ? (i / 4) | 0 : 2;+ const x = X0 + c * PW + 6, y = Y0 + r * PH, w = (wide ? 2 : 1) * PW - 12, xs = wide ? [40, 112, 184, 256] : [40], px = w - 30, py = 27;+ const g = el("g", { transform: `translate(${x} ${y})` }, svg);+ el("rect", { x: -5, y: -5, width: w + 10, height: DH + 32, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const chairs = xs.map((cx) => {+ const ch = el("g", {}, g);+ el("rect", { x: -19, y: -14, width: 38, height: 28, rx: 9, fill: col, stroke: INK, "stroke-width": 2.5 }, ch);+ el("rect", { x: -15, y: 11, width: 30, height: 8, rx: 4, fill: INK }, ch);+ return { ch, cx };+ });+ el("rect", { width: w, height: DH, rx: 10, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ xs.forEach((lx) => {+ el("rect", { x: lx - 21, y: 12, width: 42, height: 30, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: lx - 21, y: 12, width: 42, height: 11, rx: 4, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${lx - 13} 30 H${lx + 13} M${lx - 13} 36 H${lx + 13}`, stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-dasharray": "1 5" }, g);+ });+ if (!wide) {+ el("rect", { x: 76, y: 13, width: 26, height: 30, rx: 3, fill: SOFT, stroke: INK, "stroke-width": 2.5, transform: `rotate(${(i % 3) * 5 - 5} 89 28)` }, g);+ el("path", { d: "M82 22 H96 M82 28 H96 M82 34 H91", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", transform: `rotate(${(i % 3) * 5 - 5} 89 28)` }, g);+ }+ el("circle", { cx: px, cy: py, r: 12, fill: YEL, stroke: INK, "stroke-width": 2.5 }, g);+ const leaves = ANG.map(() => el("path", { d: "M0 0 Q8 -10 0 -21 Q-8 -10 0 0 Z", fill: MINT, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, g));+ el("circle", { cx: px, cy: py, r: 3, fill: INK }, g);+ const card = el("g", {}, g);+ el("rect", { x: -79, y: -17, width: 158, height: 34, rx: 6, fill: col, stroke: INK, "stroke-width": 2.5 }, card);+ fit(txt(card, 0, 6.5, name, { "text-anchor": "middle", "font-size": 18, "font-weight": 700 }), 144);+ const d = { card, leaves, chairs, cx: w / 2, cy: DH - 23, px, py, kt: -99, extra: 0 };+ button(g, line, () => { d.extra = d.extra >= 3 ? 0 : d.extra + 1; d.kt = api.now(); pinned = i; api.poke(); });+ return d;+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), idle = io(seg(t, 9.2, 10.2)) * (1 - out);+ let cur = -1;+ desks.forEach((d, i) => {+ const a = A[i], k = seg(t, a, a + 0.42), e = cl(bk(k, 1.6), 0, 1.2), tau = t - a - 0.42, w = now - d.kt;+ const rot = (1 - oc(k)) * (i % 2 ? 16 : -16) + (tau > 0 ? 3 * Math.exp(-5 * tau) * Math.sin(16 * tau) : 0);+ d.card.setAttribute("transform", `translate(${d.cx} ${(d.cy - 34 * (1 - oc(k))).toFixed(2)}) rotate(${rot.toFixed(2)}) scale(${lerp(1.9, 1, e).toFixed(3)})`);+ d.card.setAttribute("opacity", (seg(t, a, a + 0.12) * (1 - out)).toFixed(2));+ d.leaves.forEach((lf, j) => {+ const g = j < 2 ? 1 : j === 2 ? cl(bk(seg(t, a + 0.35, a + 0.95), 2.4), 0, 1.5) * (1 - out) : j - 3 >= d.extra ? 0 : j - 3 === d.extra - 1 ? cl(bk(seg(w, 0, 0.6), 2.4), 0, 1.5) : 1;+ lf.setAttribute("transform", `translate(${d.px} ${d.py}) rotate(${(ANG[j] + Math.sin(now * 1.3 + i * 0.9 + j * 2.1) * 4).toFixed(2)}) scale(${g.toFixed(3)})`);+ });+ /* one chair is left spinning */+ d.chairs.forEach(({ ch, cx }) => ch.setAttribute("transform", `translate(${cx} ${(DH + 4 + (i === 5 ? 5 * idle : 0)).toFixed(2)}) rotate(${(i === 5 ? 22 * idle * Math.sin(now * 0.9 + 1.1) : 0).toFixed(2)})`));+ if (t >= a && !out) cur = i;+ });+ const pw = pinned >= 0 ? now - desks[pinned].kt : 99;+ say(cap, pw < 4 ? SEATS[pinned][1] : cur < 0 ? "Fourteen seats, and nobody shares a job" : t >= 9.4 ? "Fourteen agents, each with one job" : SEATS[cur][1], 700);+ };+ },+ };++ /* ---- clock: the hand sweeps one day; a card slides out from behind the clock for each schedule ---- */+ SC.clock = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Which agents work on a clock?", "tap the clock");+ const CX = 400, CY = 284, R = 128;+ const hourAt = (t) => { const p = seg(t, 1, 9.4); return 24 * p * p * (3 - 2 * p); };+ const timeOf = (H) => { let a = 1, b = 9.4; for (let i = 0; i < 28; i++) { const m = (a + b) / 2; if (hourAt(m) < H) a = m; else b = m; } return b; };+ const C = [+ [152, 154, "09:00", "briefing", YEL, [9], 0, "Andy, 09:00: one briefing as a draft"],+ [152, 420, "09:30 and 21:30", "blog runs", MINT, [9.5, 21.5], 0.3, "Jim, 09:30 and 21:30: one pull request per run"],+ [648, 420, "every 6 hours", "dashboard", SKY, [6, 12, 18, 24], 0, "Ryan, every 6 hours: the company dashboard"],+ [648, 154, "every 12 hours", "task board", LIL, [12, 24], 0, "Meredith, every 12 hours: keeps the task board readable"],+ ];+ C.forEach(([x, y]) => el("rect", { x: x - 118, y: y - 52, width: 236, height: 104, rx: 14, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg));+ const cards = C.map(([x, y, a, b, col, hrs, lag, line]) => {+ const g = el("g", {}, svg);+ el("rect", { x: -118, y: -52, width: 236, height: 104, rx: 14, fill: col, stroke: INK, "stroke-width": 3 }, g);+ fit(txt(g, -98, -9, a, { "font-size": 25, "font-weight": 700 }), 196);+ txt(g, -98, 27, b, { "font-size": 22, "font-weight": 500 });+ const pips = hrs.map((H, j) => ({ c: el("circle", { cx: 100 - (hrs.length - 1 - j) * 19, cy: 20, r: 6.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g), at: timeOf(H) + lag }));+ return { g, x, y, t0: timeOf(hrs[0]) - 0.6 + lag, pips, line };+ });+ const clock = el("g", {}, svg);+ el("circle", { cx: CX, cy: CY, r: R + 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, clock);+ const pen = el("g", {}, clock);+ el("path", { d: "M0 0 V52", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, pen);+ el("circle", { cy: 52, r: 15, fill: YEL, stroke: INK, "stroke-width": 3 }, pen);+ el("circle", { cx: CX, cy: CY - R - 14, r: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, clock);+ el("circle", { cx: CX, cy: CY, r: R + 11, fill: BLUE, stroke: INK, "stroke-width": 3 }, clock);+ el("circle", { cx: CX, cy: CY, r: R - 3, fill: "#fff", stroke: INK, "stroke-width": 3 }, clock);+ /* the tick at seven was put on by eye */+ for (let i = 0; i < 12; i++) el("path", { d: `M${CX} ${CY - R + 11} V${CY - R + (i % 3 ? 25 : 37)}`, stroke: INK, "stroke-width": i % 3 ? 3 : 6, "stroke-linecap": "round", transform: `rotate(${i * 30} ${CX} ${CY}) rotate(${i === 7 ? 16 : 0} ${CX} ${CY - R + 18})` }, clock);+ const hand = el("g", {}, clock);+ el("path", { d: `M0 18 V${-(R - 44)}`, stroke: INK, "stroke-width": 9, "stroke-linecap": "round" }, hand);+ el("circle", { cy: -(R - 44), r: 9, fill: YEL, stroke: INK, "stroke-width": 3 }, hand);+ el("circle", { cx: CX, cy: CY, r: 10, fill: YEL, stroke: INK, "stroke-width": 3 }, clock);+ button(clock, "Run the day again", () => api.seek(0.9));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const tau = t - 9.4, out = seg(t, 13.2, 13.7);+ const ang = hourAt(t) * 30 + (tau > 0 ? 6 * Math.exp(-4 * tau) * Math.sin(11 * tau) : 0);+ hand.setAttribute("transform", `translate(${CX} ${CY}) rotate(${ang.toFixed(2)})`);+ pen.setAttribute("transform", `translate(${CX} ${CY + R + 8}) rotate(${(Math.sin(now * 2.6 + 0.9) * 20).toFixed(2)})`);+ let cur = -1, last = -1;+ cards.forEach((c, i) => {+ const k = seg(t, c.t0, c.t0 + 0.6), e = bk(k, 1.5);+ let bump = 0;+ c.pips.forEach((p) => {+ const w = t - p.at, on = w >= 0 && out < 1;+ p.c.setAttribute("fill", on ? INK : "#fff");+ p.c.setAttribute("r", (6.5 * (on ? 1 + 0.7 * Math.exp(-7 * w) : 1)).toFixed(2));+ if (on) bump += Math.exp(-6 * w);+ });+ tf(c.g, lerp(CX, c.x, e).toFixed(2), lerp(CY, c.y, e).toFixed(2), (lerp(0.45, 1, cl(e, 0, 1.1)) * (1 + 0.06 * bump)).toFixed(4), ((1 - oc(k)) * (i % 2 ? 10 : -10)).toFixed(2));+ c.g.setAttribute("opacity", t >= c.t0 ? (1 - out).toFixed(2) : 0);+ if (t >= c.t0 && c.t0 > last) { last = c.t0; cur = i; }+ });+ say(cap, cur < 0 || out > 0 ? "Four of ours start from a schedule" : t > 9.8 ? "Four of ours start from a schedule, with no person typing anything" : cards[cur].line, 700);+ };+ },+ };++ /* ---- gate: a drawing becomes a jigsaw piece, waits at a raised drawbridge for three checks, then joins the row ---- */+ SC.gate = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Who writes the code, and who checks it?", "tap the bridge");+ const G = 400, PIECE = "M-20 -20 H20 V-6 h5 a8 8 0 1 1 0 12 h-5 V20 H-20 V6 h5 a8 8 0 1 0 0 -12 h-5 Z";+ const piece = (p, fill) => el("path", { d: PIECE, fill, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, p);+ const water = el("path", { fill: SKY, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 30, y: G, width: 446, height: 54, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 604, y: G, width: 166, height: 54, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 118, 435, "draws first", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 });+ txt(svg, 322, 435, "build to an order", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 });+ fit(txt(svg, 687, 435, "release branch", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }), 148);+ /* drawing board */+ el("path", { d: `M70 ${G} L104 236 M166 ${G} L132 236 M118 ${G} V250`, stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, svg);+ const board = el("g", { transform: "translate(118 232) rotate(-7)" }, svg);+ el("rect", { x: -72, y: -92, width: 144, height: 170, rx: 8, fill: LIL, stroke: INK, "stroke-width": 3 }, board);+ el("rect", { x: -78, y: 78, width: 156, height: 12, rx: 5, fill: YEL, stroke: INK, "stroke-width": 3 }, board);+ el("rect", { x: -54, y: -76, width: 108, height: 140, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, board);+ el("path", { d: "M-40 -54 H40 M-40 -30 H8 M-40 -6 H24", stroke: SOFT, "stroke-width": 8, "stroke-linecap": "round" }, board);+ el("path", { d: "M22 72 H58 L68 75 L58 78 H22 Z", fill: SKY, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, board);+ /* tool rail and workbench */+ el("path", { d: "M240 146 H404 M262 146 V158 M300 146 V158 M385 146 V158", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, svg);+ el("rect", { x: 254, y: 156, width: 16, height: 96, rx: 3, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M254 172 h7 M254 188 h7 M254 204 h7 M254 220 h7 M254 236 h7", stroke: INK, "stroke-width": 2 }, svg);+ el("path", { d: "M296 156 V236 H360 Z M308 192 V224 H334 Z", fill: SKY, "fill-rule": "evenodd", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 380, y: 156, width: 10, height: 76, rx: 4, fill: LIL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 385, cy: 240, r: 11, fill: LIL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 385, cy: 240, r: 4, fill: "#fff", stroke: INK, "stroke-width": 2 }, svg);+ [254, 378].forEach((x) => el("rect", { x, y: 344, width: 12, height: 56, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg));+ el("path", { d: "M266 376 H378", stroke: INK, "stroke-width": 3 }, svg);+ const bench = el("rect", { x: 238, y: 330, width: 170, height: 15, rx: 5, fill: YEL, stroke: INK, "stroke-width": 3 }, svg);+ const ham = el("g", {}, svg);+ el("rect", { x: -50, y: -3, width: 50, height: 6, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ham);+ el("rect", { x: -62, y: -13, width: 16, height: 22, rx: 3, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, ham);+ /* the gate: a sign with three lights on a post, and the bridge it holds up */+ el("rect", { x: 462, y: 248, width: 16, height: 152, fill: BLUE, stroke: INK, "stroke-width": 3 }, svg);+ /* a lifebuoy on the post, in case a piece falls in */+ el("circle", { cx: 470, cy: 318, r: 13, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 470, cy: 318, r: 5, fill: BLUE, stroke: INK, "stroke-width": 2 }, svg);+ el("path", { d: "M470 305 V313 M470 323 V331 M457 318 H465 M475 318 H483", stroke: INK, "stroke-width": 2 }, svg);+ el("rect", { x: 462, y: 98, width: 290, height: 150, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 486, 134, "the gate", { "font-size": 26, "font-weight": 700 });+ el("path", { d: "M486 148 H728", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, svg);+ const L = [5.3, 6.1, 6.9];+ const lamps = ["reads the diff", "own tests", "full suite"].map((s, i) => {+ txt(svg, 522, 178 + i * 29, s, { "font-size": 22, "font-weight": 600 });+ return el("circle", { cx: 498, cy: 171 + i * 29, r: 10, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ });+ const br = el("g", {}, svg);+ el("rect", { x: 468, y: 258, width: 146, height: 152, rx: 12, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, br);+ const rope = el("path", { stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-dasharray": "2 6" }, br);+ const deck = el("g", {}, br);+ el("rect", { x: 0, y: -8, width: 130, height: 11, rx: 4, fill: YEL, stroke: INK, "stroke-width": 3 }, deck);+ el("path", { d: "M26 -8 V3 M52 -8 V3 M78 -8 V3 M104 -8 V3", stroke: INK, "stroke-width": 2 }, deck);+ button(br, "Send another piece to the gate", () => api.seek(0.9));+ /* the release branch: a row of pieces with room for one more */+ [[684, BLUE], [724, LIL]].forEach(([x, c]) => piece(el("g", { transform: `translate(${x} ${G - 20})` }, svg), c));+ const sp = el("g", {}, svg); star(sp, 0, 0, 13, "#fff");+ /* the sheet that leaves the board, and the piece it becomes */+ const sheet = el("g", {}, svg);+ el("rect", { x: -54, y: -70, width: 108, height: 140, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sheet);+ const parts = [+ el("rect", { x: -42, y: -58, width: 84, height: 18, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2 }, sheet),+ el("rect", { x: -42, y: -30, width: 38, height: 44, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2 }, sheet),+ el("path", { d: "M6 -24 H42 M6 -9 H42 M6 6 H28", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, sheet),+ el("rect", { x: -42, y: 28, width: 46, height: 20, rx: 10, fill: MINT, stroke: INK, "stroke-width": 2 }, sheet),+ ];+ const pc = el("g", {}, svg); piece(pc, YEL);+ const HOPS = [[4.0, 4.7, 420, G - 20, 34], [8.8, 9.3, 506, G - 28, 22], [9.3, 9.8, 586, G - 28, 22], [9.8, 10.4, 644, G - 20, 28]];+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.1, 13.5);+ let w = `M476 456 V${(424 + 3 * Math.sin(now * 2.2)).toFixed(1)}`;+ for (let x = 492; x <= 604; x += 16) w += ` L${x} ${(424 + 3 * Math.sin(x * 0.11 + now * 2.2)).toFixed(1)}`;+ water.setAttribute("d", w + " V456 Z");+ /* sheet: drawn on the board, then flies to the bench */+ parts.forEach((p, i) => p.setAttribute("opacity", seg(t, 0.25 + i * 0.26, 0.5 + i * 0.26).toFixed(2)));+ const f = io(seg(t, 1.5, 2.6)), taps = t > 2.6 && t < 3.5 ? Math.abs(Math.sin(((t - 2.6) / 0.3) * Math.PI)) : 0;+ tf(sheet, lerp(117.3, 300, f).toFixed(2), (lerp(226, 296, f) - 76 * Math.sin(Math.PI * f) + (taps ? (1 - taps) * 3 : 0)).toFixed(2), lerp(1, 0.4, f).toFixed(4), (lerp(-7, 0, f) + 16 * Math.sin(Math.PI * f)).toFixed(2));+ sheet.setAttribute("opacity", (1 - seg(t, 3.15, 3.35)).toFixed(2));+ tf(ham, 402, 323, 1, (50 * taps).toFixed(2));+ bench.setAttribute("y", (330 + (taps && taps < 0.25 ? 1.5 : 0)).toFixed(1));+ /* the piece */+ let x = 300, y = 310, r = 0;+ HOPS.forEach(([a, b, X, Y, H]) => { const p = seg(t, a, b); if (p > 0) { const e = p < 1 ? p : 1; x = lerp(x, X, e); y = lerp(y, Y, e) - H * 4 * e * (1 - e); r = 9 * Math.sin(2 * Math.PI * e); } });+ const t1 = t - 4.7, t2 = t - 10.4;+ if (t1 > 0 && t < 8.8) x -= 6 * Math.exp(-7 * t1) * Math.abs(Math.sin(13 * t1));+ if (t2 > 0) r = 5 * Math.exp(-6 * t2) * Math.sin(18 * t2);+ tf(pc, x.toFixed(2), y.toFixed(2), cl(bk(seg(t, 3.2, 3.65), 2.2), 0, 1.4).toFixed(3), r.toFixed(2));+ pc.setAttribute("opacity", (1 - out).toFixed(2));+ lamps.forEach((c, i) => { const v = t - L[i], on = v >= 0 && t < 13.2; c.setAttribute("fill", on ? MINT : SOFT); c.setAttribute("r", (10 * (on ? 1 + 0.5 * Math.exp(-7 * v) : 1)).toFixed(2)); });+ const k = io(seg(t, 7.6, 8.5)) * (1 - io(seg(t, 13.2, 13.9))), tb = t - 8.5;+ const ang = -76 * (1 - k) - (tb > 0 && t < 13.2 ? 7 * Math.exp(-5 * tb) * Math.abs(Math.sin(11 * tb)) : 0), ar = (ang * Math.PI) / 180;+ deck.setAttribute("transform", `translate(478 398) rotate(${ang.toFixed(2)})`);+ rope.setAttribute("opacity", (1 - k).toFixed(2));+ rope.setAttribute("d", `M478 258 L${(478 + 124 * Math.cos(ar)).toFixed(1)} ${(394 + 124 * Math.sin(ar)).toFixed(1)}`);+ tf(sp, 646, 334, (cl(bk(seg(t, 10.4, 10.9), 2.6), 0, 1.5) * (1 - out) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.5 || out > 0 ? "Pam draws first: every new screen is an HTML page" : t < 4 ? "Builders build to an order" : t < 5.3 ? "Kevin is the gate" : t < 10.4 ? "He reads the diff, writes his own tests, runs the full suite" : "He merges into a release branch, never into main", 700);+ };+ },+ };++ /* ---- title card, 16 by 9: a desk bell that somebody keeps ringing ---- */+ SC.hero = {+ w: 960, h: 540, dur: 3.2, still: 0.55,+ build(svg) {+ txt(svg, 66, 112, "AI agent use cases from a real company", { "font-size": 25, "font-weight": 500, fill: MUTE });+ txt(svg, 62, 204, "Every Agent", { "font-size": 82, "font-weight": 700 });+ const ti = txt(svg, 62, 294, "on Our Floor", { "font-size": 82, "font-weight": 700 });+ el("path", { d: `M68 324 H${Math.round(56 + ti.getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ el("rect", { x: 62, y: 362, width: 312, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 218, 399, "Checked 14 Oct 2026", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 });+ el("path", { d: "M70 478 C190 440 300 508 420 468 S560 470 620 452", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 478, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ const c = el("g", {}, svg), bell = el("g", {}, c);+ el("rect", { x: -7, y: -92, width: 14, height: 40, fill: "#fff", stroke: INK, "stroke-width": 3 }, bell);+ const knob = el("rect", { x: -24, y: -112, width: 48, height: 20, rx: 10, fill: LIL, stroke: INK, "stroke-width": 3 }, bell);+ el("path", { d: "M-104 72 C-104 -22 -58 -64 0 -64 C58 -64 104 -22 104 72 Z", fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, bell);+ el("path", { d: "M-70 26 C-66 -10 -44 -32 -16 -40", stroke: "#fff", "stroke-width": 9, fill: "none", "stroke-linecap": "round" }, bell);+ el("rect", { x: -124, y: 72, width: 248, height: 24, rx: 12, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ el("path", { d: "M-150 110 H150", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, c);+ const ding = el("path", { d: "M122 -46 L152 -63 M134 -4 L164 -7 M92 -86 L110 -116 M-122 -46 L-152 -63 M-134 -4 L-164 -7 M-92 -86 L-110 -116", stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, c);+ star(svg, 626, 132, 15, SKY); star(svg, 880, 468, 17, MINT); star(svg, 596, 396, 12, YEL);+ return (t) => {+ const press = Math.sin(Math.PI * seg(t, 0.1, 0.5)), tau = t - 0.3;+ knob.setAttribute("y", (-112 + 16 * press).toFixed(2));+ bell.setAttribute("transform", `rotate(${(tau > 0 ? 2.4 * Math.exp(-2.4 * tau) * Math.sin(26 * tau) : 0).toFixed(3)} 0 72)`);+ ding.setAttribute("opacity", (seg(t, 0.3, 0.42) * (1 - seg(t, 1.2, 2))).toFixed(2));+ tf(ding, 0, 0, (1 + 0.1 * oc(seg(t, 0.3, 1.2))).toFixed(3));+ tf(c, 760, 312, 0.96);+ };+ },+ };++ /* a status pill sized to its word, with or without a tick; used by the scenes below */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);++ /* ---- drafts: three clipboards; a pen signs the first two, the third only ever collects labels ---- */+ SC.drafts = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Who deals with the outside world?", "tap the pen");+ const X = [146, 400, 654], TOP = 92, COL = [YEL, SKY, LIL], NAME = ["email draft", "pull request", "triage label"], WORD = ["sent", "merged", "labels only"], BT = [4.85, 7.45, 9.1];+ const SIG = "M-56 240 C-52 220 -40 220 -42 236 C-44 248 -30 246 -26 232 C-22 220 -12 222 -14 234 C-16 246 -2 244 2 232 C6 220 16 222 14 234 C14 244 26 244 32 232 C38 220 46 236 58 230";+ const TAGS = [[-18, 152, -4, MINT, 1.7], [14, 180, 4, YEL, 2.2], [-14, 208, 2, SKY, 8.1], [20, 236, -5, BLUE, 8.6]];+ const tag = (p, col) => {+ const g = el("g", {}, p);+ el("rect", { x: -38, y: -14, width: 76, height: 28, rx: 14, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ el("circle", { cx: -26, cy: 0, r: 4.5, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ return g;+ };+ /* the desk edge the pen lies on */+ [344, 432].forEach((x) => el("path", { d: `M${x - 13} 432 V442 a13 13 0 0 0 26 0 V432`, fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg));+ el("rect", { x: 30, y: 452, width: 740, height: 40, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, 400, 479, "a person", { "text-anchor": "middle", "font-size": 20, "font-weight": 700 });+ const boards = X.map((x, i) => {+ const g = el("g", { transform: `translate(${x} ${TOP})` }, svg);+ el("rect", { x: -102, y: 0, width: 204, height: 300, rx: 14, fill: COL[i], stroke: INK, "stroke-width": 3 }, g);+ const sh = el("g", {}, g), b = { sh, tags: [] };+ el("rect", { x: -86, y: 16, width: 172, height: 268, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sh);+ txt(sh, 0, 70, NAME[i], { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ const faint = (d, col = LIL) => el("path", { d, stroke: col, "stroke-opacity": col === LIL ? 0.45 : 1, "stroke-width": 9, "stroke-linecap": "round" }, sh);+ if (i === 0) faint("M-64 100 H-8 M-64 124 H64 M-64 148 H64 M-64 172 H22");+ if (i === 1) { faint("M-64 100 H20 M-64 172 H48"); faint("M-64 124 H64 M-64 148 H30", MINT); }+ if (i === 2) {+ faint("M-38 100 H58 M-64 124 H30");+ el("circle", { cx: -57, cy: 100, r: 8, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, sh);+ b.tags = TAGS.map(([tx, ty, r, col, at]) => ({ g: tag(sh, col), tx, ty, r, at }));+ b.dang = tag(sh, MINT);+ } else {+ b.line = el("path", { d: "M-62 252 H62", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "1 8" }, sh);+ b.sig = el("path", { d: SIG, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, sh);+ }+ el("rect", { x: -40, y: -10, width: 80, height: 34, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("circle", { cx: 0, cy: 4, r: 5, fill: INK }, g);+ b.badge = pill(g, WORD[i], i === 2 ? YEL : MINT, i < 2);+ return b;+ });+ const SL = boards[0].sig.getTotalLength();+ boards.forEach((b) => b.sig && b.sig.setAttribute("stroke-dasharray", `${SL} ${SL}`));+ const at = (i, k) => { const p = boards[i].sig.getPointAtLength(cl(k) * SL); return [X[i] + p.x, TOP + p.y]; };+ const HOME = [322, 438], A0 = at(0, 0), A1 = at(0, 1), B0 = at(1, 0), B1 = at(1, 1), TILT = -55;+ const wr = (k) => lerp(k, io(k), 0.6);+ const pen = el("g", {}, svg);+ el("rect", { x: -10, y: -24, width: 156, height: 48, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, pen);+ el("rect", { x: 40, y: -12, width: 92, height: 24, rx: 11, fill: BLUE, stroke: INK, "stroke-width": 3 }, pen);+ el("path", { d: "M92 -12 V12", stroke: INK, "stroke-width": 2.5 }, pen);+ el("rect", { x: 92, y: -19, width: 36, height: 7, rx: 3.5, fill: YEL, stroke: INK, "stroke-width": 2 }, pen);+ el("rect", { x: 30, y: -11, width: 18, height: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, pen);+ el("path", { d: "M0 0 L30 -10 V10 Z", fill: YEL, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, pen);+ el("path", { d: "M8 0 H24", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, pen);+ button(pen, "Sign the drafts again", () => api.seek(0.3));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7);+ boards.forEach((b, i) => {+ const a = 0.25 + i * 0.35, k = seg(t, a, a + 0.45) * (1 - io(out)), v = t - a - 0.45;+ tf(b.sh, 0, (-24 * (1 - oc(k))).toFixed(2), 1, ((1 - oc(k)) * (i % 2 ? 6 : -6) + (v > 0 ? 2 * Math.exp(-5 * v) * Math.sin(14 * v) : 0)).toFixed(2));+ if (b.sig) {+ const s = wr(seg(t, i ? 5.5 : 3.8, i ? 6.4 : 4.7));+ b.sig.setAttribute("stroke-dashoffset", (SL * (1 - s)).toFixed(1));+ b.sig.setAttribute("opacity", s > 0 ? (1 - out).toFixed(2) : 0);+ b.line.setAttribute("opacity", s > 0 ? 1 : (0.6 + 0.4 * Math.sin(now * 4 + i * 2)).toFixed(2));+ }+ b.tags.forEach((g) => tf(g.g, g.tx, g.ty, (land(seg(t, g.at, g.at + 0.4), 2.4) * (1 - out)).toFixed(3), g.r));+ if (b.dang) {+ /* one label too many: it lets go and hangs off the edge */+ const w = t - 10, ang = w < 0 ? 172 : 98 + 74 * Math.exp(-2.2 * w) * Math.cos(5.5 * w) + 4 * Math.sin(now * 1.8) * seg(w, 1, 2);+ b.dang.setAttribute("transform", `translate(-76 150) rotate(${ang.toFixed(2)}) scale(${(land(seg(t, 9.8, 10.1), 2) * (1 - out)).toFixed(3)}) translate(26 0)`);+ }+ const bk2 = seg(t, BT[i], BT[i] + 0.45);+ tf(b.badge, 0, 300, land(bk2).toFixed(3), ((1 - oc(bk2)) * -10).toFixed(2));+ b.badge.setAttribute("opacity", bk2 > 0 ? (1 - out).toFixed(2) : 0);+ });+ let p = HOME, r = 0;+ if (t >= 3 && t < 3.8) { const k = io(seg(t, 3, 3.8)); p = [lerp(HOME[0], A0[0], k), lerp(HOME[1], A0[1], k) - 22 * Math.sin(Math.PI * k)]; r = TILT * k; }+ else if (t >= 3.8 && t < 4.7) { p = at(0, wr(seg(t, 3.8, 4.7))); r = TILT + 3 * Math.sin(t * 16); }+ else if (t >= 4.7 && t < 5.5) { const k = io(seg(t, 4.7, 5.5)); p = [lerp(A1[0], B0[0], k), lerp(A1[1], B0[1], k) - 18 * Math.sin(Math.PI * k)]; r = TILT; }+ else if (t >= 5.5 && t < 6.4) { p = at(1, wr(seg(t, 5.5, 6.4))); r = TILT + 3 * Math.sin(t * 16); }+ else if (t >= 6.4 && t < 7.4) { const k = io(seg(t, 6.55, 7.4)); p = [lerp(B1[0], HOME[0], k), lerp(B1[1] - 8 * oc(seg(t, 6.4, 6.6)), HOME[1], k)]; r = TILT * (1 - k); }+ else if (t >= 7.4) { const v = t - 7.4; r = -3.5 * Math.exp(-6 * v) * Math.sin(17 * v); }+ tf(pen, p[0].toFixed(2), p[1].toFixed(2), 1, r.toFixed(2));+ say(cap, t < 1.6 || out > 0 ? "Four agents read what is outside the office" : t < 3 ? "None of them takes the final step alone" : t < 5.4 ? "Replies to email are drafts" : t < 8 ? "Jim opens pull requests to this website. A person merges them." : "Angela can add labels. She cannot close or merge.", 700);+ };+ },+ };++ /* ---- temps: a tent is pitched for each job, reports done and is packed away; the house stays ---- */+ SC.temps = {+ dur: 15.2, still: 11.3,+ build(svg, api) {+ head(svg, "Not every job deserves a desk", "tap a job");+ const SX = [118, 290, 462], G = 440, COL = [SKY, LIL, YEL], T = [0.4, 4.1, 7.8], TENT = "M0 -165 L75 0 H-75 Z";+ const JOB = [["run the", "test suite"], ["fix one", "release bug"], ["free", "disk space"]];+ const LINE = ["Run the test suite and write an honest report", "Fix one release bug", "Free disk space on a laptop"];+ el("rect", { x: 30, y: G, width: 740, height: 46, rx: 8, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ /* the house, with its chimney going */+ const smoke = [0, 1, 2].map(() => el("circle", { fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 700, y: 196, width: 24, height: 60, fill: LIL, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M552 268 L656 176 L760 268 Z", fill: BLUE, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("rect", { x: 570, y: 268, width: 172, height: 172, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 578, y: 282, width: 156, height: 40, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ fit(txt(svg, 656, 309, "lasting agent", { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }), 140);+ el("path", { d: "M634 440 V372 a22 22 0 0 1 44 0 V440 Z", fill: YEL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, svg);+ el("circle", { cx: 668, cy: 402, r: 3.5, fill: INK }, svg);+ [584, 692].forEach((x) => {+ el("rect", { x, y: 358, width: 34, height: 40, rx: 4, fill: SKY, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: `M${x + 17} 358 V398 M${x} 378 H${x + 34}`, stroke: INK, "stroke-width": 2 }, svg);+ });+ const spots = SX.map((x, i) => {+ el("path", { d: `M${x} 186 V${G}`, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: TENT, transform: `translate(${x} ${G})`, fill: SOFT, stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg);+ const tk = el("g", { transform: `translate(${x} 150)` }, svg);+ el("rect", { x: -81, y: -46, width: 162, height: 126, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, tk);+ el("rect", { x: -75, y: -40, width: 150, height: 76, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, tk);+ el("circle", { cx: 0, cy: -40, r: 6, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, tk);+ JOB[i].forEach((s, j) => txt(tk, 0, -7 + j * 25, s, { "text-anchor": "middle", "font-size": 21, "font-weight": 700 }));+ el("rect", { x: -50, y: 50, width: 100, height: 14, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, tk);+ const bar = el("rect", { x: -47, y: 53, width: 0, height: 8, rx: 4, fill: MINT }, tk), done = pill(tk, "done", MINT, true);+ button(tk, `Start a temp to ${JOB[i].join(" ")}`, () => api.seek(T[i]));+ return { x, bar, done };+ });+ spots.forEach((s, i) => {+ const tent = (s.tent = el("g", {}, svg));+ el("path", { d: "M-46 -64 L-82 0 M46 -64 L82 0", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, tent);+ el("path", { d: TENT, fill: COL[i], stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, tent);+ el("path", { d: "M0 -112 L26 0 H-26 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, tent);+ el("path", { d: "M0 -112 V0", stroke: INK, "stroke-width": 2 }, tent);+ el("circle", { cx: 0, cy: -165, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, tent);+ });+ spots.forEach((s, i) => {+ const bag = (s.bag = el("g", {}, svg));+ el("path", { d: "M-12 -30 Q0 -47 12 -30", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, bag);+ el("rect", { x: -36, y: -30, width: 72, height: 30, rx: 13, fill: COL[i], stroke: INK, "stroke-width": 3 }, bag);+ el("path", { d: "M-14 -30 V0 M14 -30 V0", stroke: INK, "stroke-width": 2.5 }, bag);+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 14.4, 14.9);+ smoke.forEach((c, j) => {+ const p = (now * 0.3 + j / 3) % 1;+ c.setAttribute("cx", (712 + 9 * Math.sin(p * 5 + j * 2)).toFixed(1)); c.setAttribute("cy", (186 - 70 * p).toFixed(1));+ c.setAttribute("r", (5 + 9 * p).toFixed(1)); c.setAttribute("opacity", (cl(p * 6) * (1 - p * p)).toFixed(2));+ });+ let cur = -1, line = "";+ spots.forEach((s, i) => {+ /* the third job took three goes, so its tent does too */+ const u = t - T[i], D = i === 2 ? 1.45 : 0.45, w0 = 0.7 + D, w1 = w0 + 0.9, f0 = w1 + 0.6, l0 = f0 + 0.3, l1 = l0 + 0.7;+ const kin = oc(seg(u, 0, 0.7)), ko = seg(u, l0, l1), early = u < w0;+ const bx = early ? lerp(-50, s.x, kin) : lerp(s.x, -50, ko * ko * ko);+ const hop = early ? 12 * Math.abs(Math.sin(kin * Math.PI * 3)) * (1 - kin) : 0;+ const bs = u <= 0 || u >= l1 ? 0 : early ? 1 - seg(u, 0.7, 0.85) : land(seg(u, f0 + 0.1, l0), 2);+ tf(s.bag, bx.toFixed(2), (G - hop).toFixed(2), bs.toFixed(3), (early ? 8 * Math.sin(kin * Math.PI * 3) * (1 - kin) : -9 * ko).toFixed(2));+ let h = 0;+ if (u >= 0.7 && u < f0) {+ const a = 0.7;+ if (i < 2) h = bk(seg(u, a, a + D), 1.8);+ else h = u < a + 0.3 ? 0.72 * oc(seg(u, a, a + 0.3)) : u < a + 0.55 ? lerp(0.72, 0.22, io(seg(u, a + 0.3, a + 0.55))) : u < a + 0.85 ? lerp(0.22, 0.86, oc(seg(u, a + 0.55, a + 0.85))) : u < a + 1.05 ? lerp(0.86, 0.28, io(seg(u, a + 0.85, a + 1.05))) : lerp(0.28, 1, bk(seg(u, a + 1.05, a + D), 1.8));+ } else if (u >= f0 && u < f0 + 0.3) { const k = seg(u, f0, f0 + 0.3); h = 1 - k * k; }+ const jig = u > w0 && u < w1 ? 1.1 * Math.sin(u * 30) : 0, joy = u >= w1 && u < w1 + 0.35 ? 9 * Math.sin(Math.PI * seg(u, w1, w1 + 0.35)) : 0;+ s.tent.setAttribute("transform", `translate(${s.x} ${(G - joy).toFixed(2)}) rotate(${jig.toFixed(2)}) scale(${(1 + 0.2 * (1 - cl(h))).toFixed(3)} ${Math.max(h, 0.001).toFixed(3)})`);+ s.tent.setAttribute("opacity", h > 0.01 ? 1 : 0);+ s.bar.setAttribute("width", (94 * (u >= w1 ? 1 : io(seg(u, w0, w1))) * (1 - out)).toFixed(1));+ tf(s.done, 0, 57, (land(seg(u, w1, w1 + 0.4)) * (1 - out)).toFixed(3));+ if (u >= 0) { cur = i; line = u < w1 ? LINE[i] : u < l1 ? "Removed when it reports done" : ""; }+ });+ say(cap, cur < 0 || out > 0 || (cur < 2 && !line) ? "A temp is started for one job" : line || "A temp is started for one job and removed when it reports done", 700);+ };+ },+ };++ /* ---- relay: a capsule goes outbox, the app, inbox through a tube; "done" comes back and the outboxes shut ---- */+ SC.relay = {+ dur: 13.5, still: 11.4,+ build(svg, api) {+ head(svg, "How does work move between them?", "tap the outbox");+ const P1 = "M96 200 H236 C300 200 292 134 340 134 C372 134 370 200 370 252 L430 348 C430 404 480 410 564 410 H704";+ const P2 = "M704 200 H564 C500 200 508 134 460 134 C428 134 430 200 430 252 L370 348 C370 404 320 410 236 410 H96";+ [[135, LIL, 12], [665, SKY, 88]].forEach(([cx, col, tab]) => {+ el("rect", { x: cx - 95 + tab, y: 104, width: 90, height: 40, rx: 9, fill: col, stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: cx - 95, y: 124, width: 190, height: 356, rx: 16, fill: col, stroke: INK, "stroke-width": 3 }, svg);+ txt(svg, cx, 250, "outbox", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ txt(svg, cx, 460, "inbox", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ /* a drawer for the memory file */+ el("rect", { x: cx - 62, y: 280, width: 124, height: 70, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: `M${cx - 16} 315 H${cx + 16}`, stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, svg);+ });+ const tube = (d) => {+ el("path", { d, fill: "none", stroke: INK, "stroke-width": 38, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ return el("path", { d, fill: "none", stroke: "#fff", "stroke-width": 32, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ };+ const t1 = tube(P1), t2 = tube(P2), L1 = t1.getTotalLength(), L2 = t2.getTotalLength();+ const capsule = (col, word) => {+ const g = el("g", {}, svg), w = word ? 66 : 52;+ el("rect", { x: -w / 2, y: -13, width: w, height: 26, rx: 13, fill: col, stroke: INK, "stroke-width": 2.5 }, g);+ if (word) txt(g, 0, 5.5, word, { "text-anchor": "middle", "font-size": 16, "font-weight": 700 });+ else {+ el("rect", { x: -11, y: -8, width: 22, height: 16, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-6 -2 H6 M-6 3 H2", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ }+ return g;+ };+ const c1 = capsule(BLUE), c2 = capsule(MINT, "done");+ [[230, 200], [570, 200], [230, 410], [570, 410]].forEach(([x, y]) => el("rect", { x: x - 6, y: y - 23, width: 12, height: 46, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, svg));+ const flap = () => {+ const g = el("g", {}, svg);+ el("rect", { x: -5, y: 0, width: 10, height: 38, rx: 4, fill: INK }, g);+ el("circle", { r: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, g);+ return g;+ };+ const v1 = flap(), v2 = flap();+ const app = el("g", {}, svg);+ el("rect", { x: -84, y: -60, width: 168, height: 120, rx: 16, fill: YEL, stroke: INK, "stroke-width": 3 }, app);+ txt(app, 0, -6, "the app", { "text-anchor": "middle", "font-size": 26, "font-weight": 700 });+ const light = el("circle", { cx: 0, cy: 28, r: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, app);+ const tick = el("path", { d: "M-6 28 l4.5 5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, app);+ const sp = el("g", {}, svg); star(sp, 0, 0, 15, "#fff");+ const file = el("g", {}, svg);+ el("path", { d: "M-15 -19 H7 L15 -11 V19 H-15 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, file);+ el("path", { d: "M7 -19 V-11 H15 M-8 -3 H8 M-8 5 H4", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-linejoin": "round" }, file);+ const hit = el("g", {}, svg);+ el("rect", { x: 66, y: 172, width: 176, height: 90, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Send a note from the outbox", () => api.seek(0.3));+ const put = (g, path, L, s, flip, sc) => {+ const p = path.getPointAtLength(s), a = path.getPointAtLength(Math.max(0, s - 1)), b = path.getPointAtLength(Math.min(L, s + 1));+ tf(g, p.x.toFixed(2), p.y.toFixed(2), sc.toFixed(3), ((Math.atan2(b.y - a.y, b.x - a.x) * 180) / Math.PI + (flip ? 180 : 0)).toFixed(2));+ return p;+ };+ const near = (p, on) => (on && Math.abs(p.x - 400) < 80 ? cl(1 - Math.abs(p.y - 300) / 75) : 0);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 12.7, 13.2), vis = 1 - out;+ /* a small file drops into the outbox and becomes a capsule */+ const fk = seg(t, 0.9, 1.4);+ tf(file, 135, lerp(100, 200, fk * fk).toFixed(2), (land(seg(t, 0.4, 0.75), 2) * lerp(1, 0.55, fk)).toFixed(3), (-8 * Math.sin(Math.PI * fk)).toFixed(2));+ file.setAttribute("opacity", (1 - seg(t, 1.3, 1.42)).toFixed(2));+ const s1 = t < 2.4 ? 40 - 9 * Math.sin(Math.PI * seg(t, 2.05, 2.4)) : lerp(40, L1 - 40, bk(seg(t, 2.4, 4.3), 0.8));+ const s2 = t < 6.6 ? 40 - 9 * Math.sin(Math.PI * seg(t, 6.25, 6.6)) : lerp(40, L2 - 40, bk(seg(t, 6.6, 8.3), 0.8));+ const a = put(c1, t1, L1, s1, false, land(seg(t, 1.35, 1.75), 2)), b = put(c2, t2, L2, s2, true, land(seg(t, 5.6, 6), 2));+ c1.setAttribute("opacity", vis.toFixed(2)); c2.setAttribute("opacity", vis.toFixed(2));+ const pulse = Math.max(near(a, t > 2.4 && t < 4.3), near(b, t > 6.6 && t < 8.3)), shut = seg(t, 8.8, 9.2) * vis;+ tf(app, 400, 300, (1 + 0.06 * pulse).toFixed(4), (1.6 * pulse * Math.sin(t * 40)).toFixed(2));+ light.setAttribute("fill", pulse > 0.1 || shut > 0 ? MINT : "#fff");+ light.setAttribute("r", (12 * (1 + 0.35 * Math.sin(Math.PI * shut))).toFixed(2));+ tick.setAttribute("opacity", shut.toFixed(2));+ /* the thread is closed: a flap drops over each outbox */+ const vk = cl(bk(seg(t, 9, 9.5), 1.4), 0, 1.3) * (1 - io(out));+ v1.setAttribute("transform", `translate(204 181) rotate(${(180 - 180 * vk).toFixed(2)})`);+ v2.setAttribute("transform", `translate(596 181) rotate(${(180 * vk - 180).toFixed(2)})`);+ tf(sp, 400, 440, (land(seg(t, 9.4, 9.9), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 2.4 || out > 0 ? "An agent drops a small file in its outbox" : t < 5.4 ? "The app delivers it" : t < 8.8 ? "Two kinds of message, inform and done, end a thread" : "Two agents cannot reply to each other forever", 700);+ };+ },+ };++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/relay.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/relay.pngnew file mode 100644index 00000000..874e066eBinary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/relay.png differdiff --git a/docs/blog/assets/media/ai-agent-use-cases-our-office/temps.png b/docs/blog/assets/media/ai-agent-use-cases-our-office/temps.pngnew file mode 100644index 00000000..5d18d488Binary files /dev/null and b/docs/blog/assets/media/ai-agent-use-cases-our-office/temps.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.pngnew file mode 100644index 00000000..40a8464fBinary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/download.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/download.pngnew file mode 100644index 00000000..c94795a4Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/download.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/engine.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/engine.pngnew file mode 100644index 00000000..f3937b0aBinary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/engine.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fixes.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fixes.pngnew file mode 100644index 00000000..057bc795Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fixes.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fork.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fork.pngnew file mode 100644index 00000000..28ecd4deBinary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/fork.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/hero.pngindex fc51d30b..2ae7ec04 100644Binary files a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/ide.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/ide.pngnew file mode 100644index 00000000..39392202Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/ide.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.css b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.cssnew file mode 100644index 00000000..859374ca--- /dev/null+++ b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.js b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.jsnew file mode 100644index 00000000..ee26de35--- /dev/null+++ b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/motion.js@@ -0,0 +1,1055 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.++ TO ADD A SCENE: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } just above+ the line marked "ADD NEW SCENES ABOVE THIS LINE". Change nothing else in this file.+ - The drawing is 800 by 600 units (shown 1600 by 1200). dur is the loop in seconds, still is the frame for NAME.png.+ - build draws everything once and returns the frame function: t is the loop time, now never wraps (idle motion).+ - api.seek(s) jumps the loop to second s (a replay). api.now() is the clock. api.poke() redraws after a tap.+ - Helpers: el, txt, head (heading plus tap hint), strip and say (the caption line), button (tap and keyboard),+ pill, star, fit, mv (marks a group that moves, for the measurement tests), seg, io, oc, bk, land, lerp, tf. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };+ /* a status pill sized to its word, with or without a tick */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* mark a group as moving, so the measurement tests can see when it crosses a label */+ const mv = (g) => { g.setAttribute("data-mv", ""); return g; };++ const SC = {};++ /* ---- title card, 16 by 9: a shelf going up, with a spirit level on it and a screw left over ---- */+ SC.hero = {+ w: 960, h: 540, dur: 4, still: 2.6,+ build(svg) {+ txt(svg, 66, 92, "From zero to a working office", { "font-size": 25, "font-weight": 500, fill: MUTE });+ const L = ["Install", "Munder Difflin", "0.5.6"].map((s, i) => txt(svg, 62, 170 + i * 78, s, { "font-size": 76, "font-weight": 700 }));+ const w = L[1].getComputedTextLength();+ if (w > 500) L.forEach((n) => n.setAttribute("font-size", Math.floor((76 * 500) / w)));+ el("path", { d: `M68 356 H${Math.round(56 + L[2].getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ const pt = txt(svg, 0, 427, "Beginner's guide", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 }), pw = Math.round(pt.getComputedTextLength() + 64);+ svg.insertBefore(el("rect", { x: 62, y: 390, width: pw, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }), pt);+ pt.setAttribute("x", 62 + pw / 2);+ el("path", { d: "M70 494 C190 468 300 518 420 490 S560 492 620 478", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 494, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ const c = el("g", { transform: "translate(748 300)" }, svg);+ /* the pencil line on the wall, the brackets and the plank */+ el("path", { d: "M-168 40 H168", stroke: LIL, "stroke-width": 4, "stroke-linecap": "round", "stroke-dasharray": "1 13" }, c);+ [-1, 1].forEach((s) => {+ el("path", { d: `M${s * 104} 66 V140 L${s * 34} 66 Z`, fill: LIL, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, c);+ el("circle", { cx: s * 93, cy: 84, r: 4, fill: INK }, c);+ el("circle", { cx: s * 93, cy: 108, r: 4, fill: INK }, c);+ });+ el("rect", { x: -156, y: 42, width: 312, height: 26, rx: 7, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ /* the spirit level */+ const lv = el("g", { transform: "translate(-10 0)" }, c);+ el("rect", { x: -124, y: -16, width: 236, height: 58, rx: 10, fill: YEL, stroke: INK, "stroke-width": 3 }, lv);+ el("path", { d: "M-100 -16 V42 M88 -16 V42", stroke: INK, "stroke-width": 3 }, lv);+ el("rect", { x: -48, y: 0, width: 84, height: 26, rx: 13, fill: "#fff", stroke: INK, "stroke-width": 3 }, lv);+ const bubble = el("ellipse", { cy: 13, rx: 10, ry: 8, fill: MINT, stroke: INK, "stroke-width": 2.5 }, lv);+ el("path", { d: "M-19 0 V26 M7 0 V26", stroke: INK, "stroke-width": 2.5 }, lv);+ [-76, 62].forEach((x) => { el("circle", { cx: x, cy: 13, r: 10, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, lv); el("circle", { cx: x, cy: 13, r: 3.5, fill: MINT, stroke: INK, "stroke-width": 2 }, lv); });+ /* the screws that were left over, and the hex key that came in the box */+ const screw = (x, y, r) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, c);+ el("rect", { x: -11, y: -5, width: 30, height: 10, rx: 2, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-3 -5 V5 M4 -5 V5 M11 -5 V5", stroke: INK, "stroke-width": 2 }, g);+ el("rect", { x: -19, y: -10, width: 8, height: 20, rx: 3, fill: INK }, g);+ };+ screw(-74, -22, 0); screw(-128, -62, -38);+ const key = el("g", { transform: "translate(-62 -112) rotate(-16)" }, c);+ el("path", { d: "M-34 -46 V26 H36", fill: "none", stroke: INK, "stroke-width": 17, "stroke-linecap": "round", "stroke-linejoin": "round" }, key);+ el("path", { d: "M-34 -46 V26 H36", fill: "none", stroke: SKY, "stroke-width": 11, "stroke-linecap": "round", "stroke-linejoin": "round" }, key);+ /* the screwdriver, on the last screw */+ const drv = el("g", { transform: "translate(-7 0)" }, c);+ el("rect", { x: 131, y: -70, width: 8, height: 112, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, drv);+ el("rect", { x: 124, y: -80, width: 22, height: 13, rx: 3, fill: YEL, stroke: INK, "stroke-width": 2.5 }, drv);+ el("rect", { x: 115, y: -168, width: 40, height: 90, rx: 16, fill: LIL, stroke: INK, "stroke-width": 3 }, drv);+ const grip = el("path", { stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, drv);+ const turn = el("path", { d: "M100 -108 q-13 14 2 30 M170 -138 q13 14 -2 30", fill: "none", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, drv);+ star(svg, 606, 150, 15, YEL); star(svg, 886, 466, 17, MINT); star(svg, 598, 404, 12, SKY);+ return (t) => {+ bubble.setAttribute("cx", (-6 + 15 * Math.exp(-1.1 * t) * Math.cos(4.2 * t)).toFixed(2));+ const k = (t * 0.9) % 1;+ grip.setAttribute("d", [0, 1, 2].map((j) => `M${(123 + ((j + k) % 3) * 8).toFixed(1)} -152 V-96`).join(" "));+ turn.setAttribute("transform", `translate(0 ${(3 * Math.sin(t * 5)).toFixed(2)})`);+ };+ },+ };++ /* ---- office: a job slip lands on the front desk, is torn in three, and one piece comes back as a decision ---- */+ SC.office = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "What is Munder Difflin?", "tap the front desk");+ const BX = [118, 306, 494, 682], COL = [SKY, LIL, MINT, BLUE], SY = 172, FX = 400, FY = 354;+ const pp = (v, L) => L - Math.abs((((v % (2 * L)) + 2 * L) % (2 * L)) - L);+ el("path", { d: "M60 130 V108 H306 M494 108 H740 V130", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round", "stroke-dasharray": "1 9" }, svg);+ txt(svg, 400, 116, "other agents", { "text-anchor": "middle", "font-size": 24, "font-weight": 700, fill: MUTE });+ el("path", { d: "M44 492 H756", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ /* a filing cabinet and a water cooler, because it is an office */+ el("rect", { x: 62, y: 376, width: 108, height: 116, rx: 6, fill: LIL, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M62 434 H170", stroke: INK, "stroke-width": 3 }, svg);+ [398, 456].forEach((y) => el("rect", { x: 98, y, width: 36, height: 10, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 662, y: 318, width: 52, height: 86, rx: 16, fill: SKY, stroke: INK, "stroke-width": 3 }, svg);+ const bub = [0, 1].map(() => el("circle", { fill: "#fff", stroke: INK, "stroke-width": 2 }, svg));+ el("rect", { x: 652, y: 398, width: 72, height: 94, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: 678, y: 420, width: 20, height: 12, rx: 3, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M670 452 H706", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, svg);+ /* the back row: four desks on a step, each with a screen */+ el("rect", { x: 30, y: 270, width: 740, height: 20, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, svg);+ const desks = BX.map((x, i) => {+ el("rect", { x: x - 62, y: 228, width: 124, height: 43, fill: COL[i], stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: x - 74, y: 216, width: 148, height: 13, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("rect", { x: x - 46, y: 243, width: 92, height: 14, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const bar = el("rect", { x: x - 43, y: 246, width: 0, height: 8, rx: 4, fill: MINT }, svg);+ const mon = el("g", {}, svg);+ el("rect", { x: -7, y: -2, width: 14, height: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, mon);+ el("rect", { x: -50, y: -72, width: 100, height: 72, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, mon);+ /* the desk with no work has found the screensaver */+ const saver = i === 3 ? el("rect", { width: 18, height: 18, rx: 5, fill: LIL, stroke: INK, "stroke-width": 2.5 }, mon) : null;+ el("rect", { x: x - 26, y: 210, width: 52, height: 7, rx: 3, fill: INK }, svg);+ const tick = el("g", {}, svg);+ el("circle", { r: 13, fill: MINT, stroke: INK, "stroke-width": 2.5 }, tick);+ el("path", { d: "M-6 0 l4.5 5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, tick);+ return { x, bar, mon, saver, tick };+ });+ /* the front desk */+ const front = el("g", {}, svg);+ el("rect", { x: 192, y: 296, width: 416, height: 204, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, front);+ el("rect", { x: 220, y: 434, width: 360, height: 58, fill: YEL, stroke: INK, "stroke-width": 3 }, front);+ el("rect", { x: 200, y: 420, width: 400, height: 15, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, front);+ txt(front, 400, 473, "your orchestrator", { "text-anchor": "middle", "font-size": 27, "font-weight": 700 });+ const fmon = el("g", {}, front);+ el("rect", { x: -9, y: -2, width: 18, height: 14, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, fmon);+ el("rect", { x: -100, y: -100, width: 200, height: 100, rx: 9, fill: "#fff", stroke: INK, "stroke-width": 3 }, fmon);+ el("rect", { x: 364, y: 414, width: 72, height: 7, rx: 3, fill: INK }, front);+ button(front, "Send another job to the front desk", () => api.seek(0.3));+ /* the slips */+ const J = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -62, y: -36, width: 124, height: 72, rx: 7, fill: SOFT, stroke: INK, "stroke-width": 3 }, J);+ const jt = txt(J, 0, 9, "your job", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 });+ const perf = el("path", { d: "M-21 -30 V30 M21 -30 V30", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "2 7" }, J);+ const S = [SKY, LIL, MINT].map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ const r = el("rect", { x: -20, y: -36, width: 40, height: 72, rx: 6, fill: c, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-10 -16 H10 M-10 -2 H10 M-10 12 H4", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, r, c };+ });+ const D = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -80, y: -40, width: 160, height: 80, rx: 8, fill: YEL, stroke: INK, "stroke-width": 3 }, D);+ txt(D, 0, -5, "needs your", { "text-anchor": "middle", "font-size": 24, "font-weight": 700 });+ txt(D, 0, 24, "decision", { "text-anchor": "middle", "font-size": 24, "font-weight": 700 });+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 14, "#fff");+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out;+ bub.forEach((c, j) => {+ const p = (now * 0.3 + j / 2) % 1;+ c.setAttribute("cx", (688 + 8 * Math.sin(p * 6 + j * 2.4)).toFixed(1)); c.setAttribute("cy", (392 - 56 * p).toFixed(1));+ c.setAttribute("r", (2.5 + 3.5 * p).toFixed(1)); c.setAttribute("opacity", (cl(p * 6) * (1 - p * p)).toFixed(2));+ });+ /* the job arrives and is torn in three */+ const h = io(seg(t, 1.2, 2.1)), shake = t > 2.7 && t < 3 ? 3 * Math.sin(t * 70) : 0;+ tf(J, (lerp(116, FX, h) + shake).toFixed(2), (lerp(338, FY, h) - 40 * Math.sin(Math.PI * h)).toFixed(2), land(seg(t, 0.25, 0.65), 2).toFixed(3), ((1 - h) * -4 + 10 * Math.sin(Math.PI * h)).toFixed(2));+ J.setAttribute("opacity", (1 - seg(t, 3, 3.1)).toFixed(2));+ jt.setAttribute("opacity", (1 - seg(t, 2.25, 2.45)).toFixed(2));+ perf.setAttribute("opacity", seg(t, 2.45, 2.65).toFixed(2));+ S.forEach((s, i) => {+ const a = 3.6 + 0.4 * i, k = io(seg(t, a, a + 0.9)), fan = oc(seg(t, 3, 3.4)), stuck = i === 1 && t > 6.6;+ tf(s.g, lerp(FX + (i - 1) * lerp(41, 58, fan), BX[i], k).toFixed(2), (lerp(FY, SY, k) - 34 * Math.sin(Math.PI * k)).toFixed(2), lerp(1, 0.8, k).toFixed(3), ((i - 1) * 7 * fan * (1 - k) + (stuck ? 5 * Math.sin(t * 13) : 0)).toFixed(2));+ s.r.setAttribute("fill", stuck ? YEL : s.c);+ s.g.setAttribute("opacity", t < 3 ? 0 : i === 1 ? (1 - seg(t, 8.8, 8.95)).toFixed(2) : vis.toFixed(2));+ });+ desks.forEach((d, i) => {+ const tau = t - (4.5 + 0.4 * i);+ d.mon.setAttribute("transform", `translate(${d.x} 208) rotate(${(i < 3 && tau > 0 ? 4 * Math.exp(-6 * tau) * Math.sin(18 * tau) : 0).toFixed(2)})`);+ const p = i === 0 ? io(seg(t, 4.7, 7)) : i === 1 ? 0.55 * oc(seg(t, 5.1, 6.4)) : i === 2 ? io(seg(t, 5.5, 8.2)) : 0;+ d.bar.setAttribute("width", (86 * p * vis).toFixed(1));+ d.bar.setAttribute("fill", i === 1 && t > 6.6 ? YEL : MINT);+ const T = i === 0 ? 7 : i === 2 ? 8.2 : 99;+ tf(d.tick, d.x + 50, 136, (land(seg(t, T, T + 0.4), 2.4) * vis).toFixed(3));+ if (d.saver) { d.saver.setAttribute("x", (-41 + pp(now * 30 + 22, 64)).toFixed(1)); d.saver.setAttribute("y", (-64 + pp(now * 23 + 9, 38)).toFixed(1)); }+ });+ /* one piece comes back */+ const e = io(seg(t, 8.8, 9.8)), tau = t - 9.8, w1 = t - 2.1, wob = (v) => (v > 0 ? Math.exp(-6 * v) * Math.sin(17 * v) : 0);+ tf(D, lerp(BX[1], FX, e).toFixed(2), (lerp(SY, FY, e) - 30 * Math.sin(Math.PI * e)).toFixed(2), (lerp(0.22, 1, e) * (1 + (tau > 0 ? 0.1 * Math.exp(-5 * tau) * Math.sin(15 * tau) : 0) + 0.02 * seg(t, 10.6, 11.4) * Math.sin(now * 3))).toFixed(4), ((1 - e) * -12 + 3 * wob(tau)).toFixed(2));+ D.setAttribute("opacity", t >= 8.8 ? vis.toFixed(2) : 0);+ fmon.setAttribute("transform", `translate(400 404) rotate(${(2.2 * (wob(w1) + wob(tau))).toFixed(2)})`);+ tf(sp, 532, 316, (land(seg(t, 9.8, 10.3), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.2 || out > 0 ? "A small office of AI workers" : t < 2.6 ? "You brief one orchestrator" : t < 3.6 ? "He splits it into pieces" : t < 8.8 ? "He hands the pieces to other agents" : "Only the decisions come back to you", 700);+ };+ },+ };++ /* ---- rooms: a house plan where every room is a folder; a new room is built and the agent's desk goes only there ---- */+ SC.rooms = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "What can it touch on your computer?", "tap a room");+ const R = [+ { n: "Documents", x: 44, y: 96, w: 236, h: 180, c: SKY, dx: 226, dy: 226 },+ { n: "Photos", x: 280, y: 96, w: 238, h: 180, c: YEL, dx: 466, dy: 226 },+ { n: "Projects", x: 518, y: 96, w: 238, h: 180, c: SKY, dx: 704, dy: 226 },+ { n: "Downloads", x: 44, y: 276, w: 256, h: 214, c: YEL, dx: 246, dy: 408 },+ ];+ const NX = 300, NY = 276, NW = 456, NH = 214, FOLD = "M0 -11 H11 L16 -6 H30 V11 H0 Z", NEW = "M300 490 H756 V276";+ const given = [false, false, false, false], gt = [-99, -99, -99, -99];+ R.forEach((r) => {+ el("rect", { x: r.x, y: r.y, width: r.w, height: r.h, fill: SOFT }, svg);+ r.fl = el("rect", { x: r.x, y: r.y, width: r.w, height: r.h, fill: r.c, opacity: 0 }, svg);+ r.in = el("g", {}, svg);+ });+ /* the plot where the new room will go: hatched, until it has a floor */+ const hatch = el("path", { d: [0, 1, 2, 3, 4, 5, 6].map((k) => `M${318 + 48 * k} 480 l136 -136`).join(" "), stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "6 9" }, svg);+ const nfl = el("rect", { x: NX, y: NY, width: NW, height: NH, fill: MINT, opacity: 0 }, svg);+ const sheet = (p, x, y, r) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, p);+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-11 -12 H11 M-11 -2 H11 M-11 8 H5", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ return g;+ };+ /* Documents: two tidy stacks. Photos: two prints. Projects: two folders. */+ [88, 146].forEach((x) => [8, 4, 0].forEach((o) => sheet(R[0].in, x + o, 224 - o, 0)));+ [[330, 224, -7], [390, 218, 5]].forEach(([x, y, r]) => {+ const g = el("g", { transform: `translate(${x} ${y}) rotate(${r})` }, R[1].in);+ el("rect", { x: -27, y: -23, width: 54, height: 46, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -20, y: -16, width: 40, height: 26, fill: SKY, stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-20 10 L-7 -5 L2 4 L8 -2 L20 10 Z", fill: MINT, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, g);+ el("circle", { cx: 11, cy: -8, r: 4, fill: YEL, stroke: INK, "stroke-width": 2 }, g);+ });+ [[540, LIL], [598, MINT]].forEach(([x, c]) => el("path", { d: "M0 -18 H16 L23 -11 H46 V20 H0 Z", fill: c, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} 222)` }, R[2].in));+ /* Downloads: a heap, and it keeps growing */+ [[-32, -18, -22], [26, -22, 14], [-8, 4, 38], [32, 16, -9], [-34, 22, 8], [4, -30, -31], [-18, 28, -47], [30, -4, 62], [0, 6, -4]].forEach(([dx, dy, r]) => sheet(R[3].in, 122 + dx, 410 + dy, r));+ const drops = [[14, -14, 21, 3.2], [-24, 2, -15, 7], [20, 22, 33, 10.4]].map(([dx, dy, r, at]) => ({ g: sheet(R[3].in, 0, 0, 0), x: 122 + dx, y: 410 + dy, r, at }));+ /* walls: the plot for the new room is dotted until it is built */+ el("path", { d: NEW, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, svg);+ const nw = el("path", { d: NEW, fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round", "stroke-dasharray": "670 670" }, svg);+ el("path", { d: "M44 96 H756 V276 H44 Z M280 96 V276 M518 96 V276 M44 276 V490 H300 V276", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ const door = (p, gap, leaf) => { el("path", { d: gap, stroke: "#fff", "stroke-width": 8 }, p); el("path", { d: leaf, fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, p); };+ door(svg, "M280 156 V192", "M280 192 H244 A36 36 0 0 1 280 156");+ door(svg, "M518 156 V192", "M518 192 H482 A36 36 0 0 1 518 156");+ const nd = el("g", {}, svg); door(nd, "M300 438 V478", "M300 478 H340 A40 40 0 0 0 300 438");+ /* each room is labelled as a folder */+ R.forEach((r) => {+ r.ic = el("path", { d: FOLD, fill: "#fff", stroke: MUTE, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${r.x + 18} ${r.y + 34})` }, svg);+ r.tx = txt(svg, r.x + 56, r.y + 43, r.n, { "font-size": 25, "font-weight": 700, fill: MUTE });+ });+ const plus = el("g", {}, svg);+ el("circle", { r: 36, fill: "#fff", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, plus);+ el("path", { d: "M-15 0 H15 M0 -15 V15", stroke: LIL, "stroke-width": 6, "stroke-linecap": "round" }, plus);+ const nl = el("g", {}, svg);+ const ni = el("path", { d: FOLD, fill: "#fff", stroke: MUTE, "stroke-width": 2.5, "stroke-linejoin": "round", transform: "translate(0 -9)" }, nl);+ const nt = txt(nl, 40, 0, "new empty folder", { "font-size": 26, "font-weight": 700, fill: MUTE });+ const files = [[356, 6, -5], [414, 7.2, 3], [472, 8.4, -2]].map(([x, at, r]) => ({ g: mv(sheet(svg, 0, 0, 0)), x, at, r }));+ /* a desk seen from above, with a small terminal on it */+ const desk = () => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -14, y: 14, width: 28, height: 24, rx: 9, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -42, y: -24, width: 84, height: 46, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: -22, y: -16, width: 44, height: 30, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: -22, y: -16, width: 44, height: 9, rx: 4, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-14 -2 l5 4 l-5 4", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return { g, cur: el("path", { d: "M-3 6 H8", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g) };+ };+ R.forEach((r) => (r.desk = desk()));+ const ad = desk(), tag = mv(pill(svg, "agent works here", YEL));+ tag.setAttribute("pointer-events", "none");+ R.forEach((r, i) => {+ const g = el("g", {}, svg);+ el("rect", { x: r.x + 9, y: r.y + 9, width: r.w - 18, height: r.h - 18, rx: 10, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ button(g, `Give the agent the ${r.n} folder`, () => { given[i] = !given[i]; gt[i] = api.now(); api.poke(); });+ });+ const ng = el("g", {}, svg);+ el("rect", { x: NX + 9, y: NY + 9, width: NW - 18, height: NH - 18, rx: 10, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, ng);+ button(ng, "Make the new empty folder again", () => api.seek(1.2));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, blink = Math.sin(now * 6) > 0 ? 1 : 0;+ /* the new room is built */+ const built = io(seg(t, 1.6, 2.5)) * (1 - io(out));+ nw.setAttribute("stroke-dashoffset", (670 * (1 - built)).toFixed(1));+ nw.setAttribute("opacity", built > 0.002 ? 1 : 0);+ nfl.setAttribute("opacity", (seg(t, 2.2, 2.9) * vis).toFixed(2));+ hatch.setAttribute("opacity", (1 - seg(t, 1.9, 2.3) * (1 - seg(t, 13.5, 13.8))).toFixed(2));+ nd.setAttribute("opacity", (seg(t, 2.6, 2.9) * vis).toFixed(2));+ const made = t >= 2.4 && t < 13.4;+ tf(nl, 320, 320, (1 + 0.1 * Math.sin(Math.PI * seg(t, 2.4, 2.85))).toFixed(3));+ nt.setAttribute("fill", made ? INK : MUTE); ni.setAttribute("stroke", made ? INK : MUTE);+ tf(plus, 528, 412, ((1 - oc(seg(t, 1.5, 1.9)) + oc(seg(t, 13.4, 13.8))) * (1 + 0.05 * Math.sin(now * 3))).toFixed(3));+ /* the desk moves in, and work starts */+ const dk = seg(t, 3.6, 4.2);+ tf(ad.g, 630, (380 - 30 * (1 - oc(dk))).toFixed(2), (lerp(1.5, 1, cl(bk(dk, 2), 0, 1.3)) * (dk > 0 ? vis : 0)).toFixed(3), ((1 - oc(dk)) * -10).toFixed(2));+ ad.cur.setAttribute("opacity", blink);+ const tk = seg(t, 4.5, 4.95);+ tf(tag, 630, 448, (land(tk) * vis).toFixed(3), ((1 - oc(tk)) * -8).toFixed(2));+ files.forEach((f) => tf(f.g, f.x, 400, (land(seg(t, f.at, f.at + 0.4), 2.4) * vis).toFixed(3), f.r));+ drops.forEach((d) => {+ const k = seg(t, d.at, d.at + 0.45), w = t - d.at - 0.45;+ tf(d.g, d.x, d.y, (k > 0 ? lerp(1.7, 1, cl(bk(k, 1.6), 0, 1.2)) * vis : 0).toFixed(3), (d.r + (1 - oc(k)) * 40 + (w > 0 ? 6 * Math.exp(-5 * w) * Math.sin(15 * w) : 0)).toFixed(2));+ });+ /* any room you tap is given to the agent as well */+ let fresh = false;+ R.forEach((r, i) => {+ const w = now - gt[i], g = given[i] ? oc(seg(w, 0, 0.45)) : 1 - oc(seg(w, 0, 0.35));+ r.fl.setAttribute("opacity", g.toFixed(2));+ r.in.setAttribute("opacity", lerp(0.42, 1, g).toFixed(2));+ r.tx.setAttribute("fill", g > 0.5 ? INK : MUTE); r.ic.setAttribute("stroke", g > 0.5 ? INK : MUTE);+ tf(r.desk.g, r.dx, r.dy, (given[i] ? land(seg(w, 0.12, 0.6), 2.2) : 1 - seg(w, 0, 0.2)).toFixed(3), (given[i] ? (1 - oc(seg(w, 0.12, 0.6))) * -10 : 0).toFixed(2));+ r.desk.cur.setAttribute("opacity", blink);+ if (given[i] && w < 4.5) fresh = true;+ });+ say(cap, fresh ? "Agents work inside the folders you give them" : t < 1.6 || out > 0 ? "You choose the folders" : t < 3.6 ? "For your first job, make a new empty folder" : "An agent works only in the folders you give it", 700);+ };+ },+ };++ /* ---- engine: three cards on a table, one per thing you might pay for; each turns over to show its engine ---- */+ SC.engine = {+ dur: 14, still: 11.5,+ build(svg, api) {+ head(svg, "Which AI engine should you pick?", "tap a card");+ const X = [150, 400, 650], CY = 284, COL = [YEL, SKY, MINT], REST = [-2.5, 1.5, -1.5], F = [2, 4, 6], HOLD = 4.5;+ const PAY = ["Claude", "ChatGPT", "nothing"], WITH = [["with a Claude", "subscription"], ["with a ChatGPT", "subscription"], ["if you pay", "for nothing"]];+ const mid = (a) => Object.assign({ "text-anchor": "middle" }, a);+ const mats = X.map(() => el("rect", { x: -114, y: -176, width: 228, height: 352, rx: 24, fill: SOFT }, svg));+ const cards = X.map((x, i) => {+ const g = mv(el("g", {}, svg));+ el("rect", { x: -111, y: -173, width: 222, height: 346, rx: 22, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const fr = el("g", {}, g), ba = el("g", {}, g);+ /* the side that asks */+ el("rect", { x: -106, y: -168, width: 212, height: 336, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 3 }, fr);+ el("circle", { cx: -78, cy: -140, r: 11, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, fr);+ el("circle", { cx: 78, cy: 140, r: 11, fill: COL[i], stroke: INK, "stroke-width": 2.5 }, fr);+ txt(fr, 0, -18, "You pay for", mid({ "font-size": 25, "font-weight": 500 }));+ const pw = fit(txt(fr, 0, 34, PAY[i], mid({ "font-size": 46, "font-weight": 700 })), 184).getComputedTextLength();+ el("path", { d: `M${(-pw / 2 + 4).toFixed(1)} 58 H${(pw / 2 - 4).toFixed(1)}`, stroke: COL[i], "stroke-width": 10, "stroke-linecap": "round" }, fr);+ /* the side that answers */+ el("rect", { x: -106, y: -168, width: 212, height: 336, rx: 18, fill: COL[i], stroke: INK, "stroke-width": 3 }, ba);+ el("circle", { cx: -78, cy: -140, r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ba);+ el("circle", { cx: 78, cy: 140, r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ba);+ txt(ba, 0, -64, "use", mid({ "font-size": 22, "font-weight": 500 }));+ fit(txt(ba, 0, -22, ["Claude Code", "Codex", "Antigravity"][i], mid({ "font-size": 32, "font-weight": 700 })), 188);+ el("path", { d: "M-74 8 H74", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, ba);+ WITH[i].forEach((s, j) => txt(ba, 0, 56 + j * 30, s, mid({ "font-size": 22, "font-weight": 600 })));+ return { g, fr, ba };+ });+ const st = X.map(() => ({ a: 0, m: null }));+ cards.forEach((c, i) => button(c.g, `Turn over the card for ${PAY[i]}`, () => {+ const back = Math.cos(Math.PI * st[i].a) < 0;+ st[i].m = { at: api.now(), a0: back ? 1 : 0, a1: back ? 0 : 1 };+ api.poke();+ }));+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const end = seg(t, 12.9, 13.1);+ cards.forEach((c, i) => {+ const s = st[i], kin = seg(t, 0.1 + 0.18 * i, 0.6 + 0.18 * i), ku = seg(t, F[i], F[i] + (i === 2 ? 1.5 : 0.8)), kd = seg(t, 13.1 + 0.1 * i, 13.6 + 0.1 * i);+ /* the card that costs nothing turns over once more than it needs to, and lands crooked */+ let a = (i === 2 ? 3 : 1) * io(ku) - io(kd), lift = (i === 2 ? 1.4 : 1) * Math.sin(Math.PI * ku) + Math.sin(Math.PI * kd), crook = i === 2 ? 5 * io(seg(t, 6.7, 7.5)) * (1 - cl(bk(seg(t, 8.1, 8.55), 1.8), 0, 1.2)) : 0;+ if (s.m) {+ const w = now - s.m.at, goal = Math.cos(Math.PI * a) < 0 ? 1 : 0;+ if (w < HOLD) { const k = seg(w, 0, 0.6); a = lerp(s.m.a0, s.m.a1, io(k)); lift = Math.sin(Math.PI * k); crook = 0; }+ else if (goal === s.m.a1 || w > HOLD + 0.6) s.m = null;+ else { const k = seg(w - HOLD, 0, 0.6); a = lerp(s.m.a1, goal, io(k)); lift = Math.sin(Math.PI * k); crook = 0; }+ }+ s.a = a;+ const cs = Math.cos(Math.PI * a), back = cs < 0, sx = Math.max(Math.abs(cs), 0.02), sc = 1 + 0.07 * lift;+ const rot = REST[i] + (1 - oc(kin)) * (i % 2 ? 3 : -3) + crook + 0.6 * Math.sin(now * 1.1 + i * 2.1) * seg(t, 8.8, 9.6) * (1 - end);+ const y = CY + 26 * (1 - oc(kin)) - 16 * lift;+ c.g.setAttribute("transform", `translate(${X[i]} ${y.toFixed(2)}) rotate(${rot.toFixed(2)}) scale(${(sx * sc).toFixed(4)} ${sc.toFixed(4)})`);+ c.fr.setAttribute("opacity", back ? 0 : 1); c.ba.setAttribute("opacity", back ? 1 : 0);+ mats[i].setAttribute("transform", `translate(${X[i]} ${(CY + 26 * (1 - oc(kin))).toFixed(2)}) rotate(${REST[i]}) scale(${(1 + 0.03 * lift).toFixed(4)})`);+ });+ say(cap, t < 2 || t > 13.1 ? "What you already pay for usually decides it" : t < 4 ? "You pay for Claude: use Claude Code" : t < 6 ? "You pay for ChatGPT: use Codex" : t < 8.7 ? "You pay for nothing: use Antigravity" : "You need exactly one to start", 700);+ };+ },+ };++ /* ---- terminal: a light text window above a row of keys; three lines are typed and each one is sent with Enter ---- */+ SC.terminal = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Install an engine and sign in", "tap the terminal");+ const CMD = ["npm install -g @openai/codex", "codex", "codex --version"];+ const X0 = 92, BY = [160, 228, 296, 362], TY = [0.7, 5.7, 9.5], TD = [2, 0.5, 1.1], ENT = [2.9, 6.4, 10.8], PR = [0, 5.45, 9.25, 11.7];+ const replay = () => api.seek(0.3);+ const tickDot = (p, r) => {+ const g = el("g", {}, p);+ el("circle", { r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${(-0.46 * r).toFixed(1)} ${(0.04 * r).toFixed(1)} l${(0.34 * r).toFixed(1)} ${(0.36 * r).toFixed(1)} l${(0.6 * r).toFixed(1)} ${(-0.76 * r).toFixed(1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* the keys: seven plain ones and Enter */+ const keys = [0, 1, 2, 3, 4, 5, 6].map((k) => el("rect", { x: 44 + 78 * k, y: 404, width: 68, height: 64, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg));+ const ek = el("g", {}, svg);+ el("rect", { x: 590, y: 404, width: 166, height: 64, rx: 12, fill: YEL, stroke: INK, "stroke-width": 3 }, ek);+ txt(ek, 652, 445, "Enter", { "text-anchor": "middle", "font-size": 26, "font-weight": 700 });+ el("path", { d: "M728 424 V438 H702 M710 430 l-8 8 l8 8", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, ek);+ ek.classList.add("mg-hit"); ek.addEventListener("click", replay);+ /* the window: a white title bar and a soft page, ink type */+ const term = el("g", {}, svg);+ el("rect", { x: 37, y: 85, width: 726, height: 306, rx: 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, term);+ el("rect", { x: 44, y: 92, width: 712, height: 292, rx: 14, fill: "#fff" }, term);+ el("path", { d: "M44 130 H756 V370 a14 14 0 0 1 -14 14 H58 a14 14 0 0 1 -14 -14 Z", fill: SOFT }, term);+ el("path", { d: "M44 130 H756", stroke: INK, "stroke-width": 3 }, term);+ el("rect", { x: 44, y: 92, width: 712, height: 292, rx: 14, fill: "none", stroke: INK, "stroke-width": 3 }, term);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 20 * i, cy: 111, r: 6, fill: c, stroke: INK, "stroke-width": 2.5 }, term));+ txt(term, 400, 118, "Terminal", { "text-anchor": "middle", "font-size": 20, fill: MUTE });+ const chev = BY.map((y) => el("path", { d: `M70 ${y - 14} l8 7 l-8 7`, fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, term));+ const body = el("g", {}, term);+ const line = CMD.map((s, i) => { const n = txt(body, X0, BY[i], s, { "font-family": MONO, "font-size": 22, "font-weight": 700 }); n.style.whiteSpace = "pre"; return n; });+ const CW = line[0].getComputedTextLength() / CMD[0].length;+ const cur = el("rect", { y: -19, width: 12, height: 24, rx: 2, fill: INK }, term);+ /* what each line prints: a progress bar, a wait for the browser, a version */+ const bar = el("g", {}, body);+ const blk = Array.from({ length: 12 }, (_, j) => el("rect", { x: X0 + 29 * j, y: 176, width: 24, height: 16, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, bar));+ const wait = [0, 1, 2].map((j) => el("circle", { cx: 98 + 16 * j, cy: 252, r: 4.5, fill: INK }, body));+ const arrow = el("path", { fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-dasharray": "1 9" }, body);+ const ahead = el("path", { d: "M-9 -8 L0 0 L-9 8", fill: "none", stroke: BLUE, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, body);+ const ver = [0, 1, 2].map(() => el("rect", { x: -7.5, y: -9, width: 15, height: 18, rx: 4, fill: INK }, body));+ const vdot = [0, 1].map((j) => el("circle", { cx: X0 + 20 + 25 * j, cy: 327, r: 2.3, fill: INK }, body));+ const vt = txt(body, 202, 327, "a version number", { "font-size": 20 });+ const tk = [0, 1, 2].map(() => tickDot(body, 11));+ /* the browser that the second line opens */+ const B = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -108, y: -98, width: 216, height: 196, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, B);+ el("path", { d: "M-108 -64 H108", stroke: INK, "stroke-width": 3 }, B);+ [0, 1, 2].forEach((j) => el("circle", { cx: -90 + 16 * j, cy: -81, r: 5, fill: SOFT, stroke: INK, "stroke-width": 2 }, B));+ el("rect", { x: -36, y: -91, width: 128, height: 20, rx: 10, fill: SOFT }, B);+ txt(B, 0, -16, "Sign in", { "text-anchor": "middle", "font-size": 28, "font-weight": 700 });+ const sb = el("g", {}, B);+ el("rect", { x: -62, y: -20, width: 124, height: 40, rx: 20, fill: YEL, stroke: INK, "stroke-width": 3 }, sb);+ el("path", { d: "M-16 0 H14 M6 -8 l8 8 l-8 8", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, sb);+ const bt = tickDot(B, 24);+ button(term, "Run the three terminal steps again", replay);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, on = Math.sin(now * 6) >= 0 ? 1 : 0;+ body.setAttribute("opacity", vis.toFixed(2));+ /* typing: one key goes down for each letter, then Enter */+ let row = -1, col = 0, typing = false, key = -1, ph = 0;+ CMD.forEach((s, i) => {+ const k = seg(t, TY[i], TY[i] + TD[i]), n = Math.min(s.length, Math.floor(k * s.length + 1e-6)), shown = s.slice(0, n);+ if (line[i]._s !== shown) { line[i]._s = shown; line[i].textContent = shown; }+ if (t >= PR[i] && t < ENT[i] + 0.1) { row = i; col = n; if (k > 0 && k < 1) { typing = true; key = (n * 3 + i * 2 + 1) % 7; ph = k * s.length - n; } }+ });+ if (t >= PR[3] && t < 13.2) { row = 3; col = 0; }+ if (t >= 13.7) { row = 0; col = 0; }+ chev.forEach((c, i) => c.setAttribute("opacity", t >= PR[i] ? (i ? vis.toFixed(2) : 1) : 0));+ cur.setAttribute("transform", `translate(${(X0 + col * CW + (col ? 3 : 0)).toFixed(1)} ${BY[Math.max(row, 0)]})`);+ cur.setAttribute("opacity", row < 0 ? 0 : typing ? 1 : on);+ keys.forEach((r, k) => { const p = k === key ? Math.sin(Math.PI * ph) : 0; r.setAttribute("y", (404 + 5 * p).toFixed(2)); r.setAttribute("fill", p > 0.25 ? LIL : "#fff"); });+ const ep = Math.max(...ENT.map((e) => Math.sin(Math.PI * seg(t, e - 0.04, e + 0.26))));+ ek.setAttribute("transform", `translate(${(673 * 0.04 * ep).toFixed(2)} ${(6 * ep + 436 * 0.04 * ep).toFixed(2)}) scale(${(1 - 0.04 * ep).toFixed(4)})`);+ /* line one: the bar rushes to the last block, loses one, then finishes */+ const nb = t < 3.1 ? 0 : t < 3.85 ? Math.floor(11 * oc(seg(t, 3.1, 3.8)) + 0.001) : t < 4.45 ? 11 : t < 5.05 ? 10 : 12;+ bar.setAttribute("opacity", t >= ENT[0] + 0.15 ? 1 : 0);+ bar.setAttribute("transform", `translate(${(t > 4.45 && t < 4.8 ? 2.5 * Math.sin((t - 4.45) * 55) * (1 - seg(t, 4.45, 4.8)) : 0).toFixed(2)} 0)`);+ blk.forEach((r, j) => r.setAttribute("fill", j < nb ? MINT : "#fff"));+ tf(tk[0], 452, 184, land(seg(t, 5.1, 5.5), 2.4).toFixed(3));+ /* line two: the browser slides in, you sign in there */+ const ak = io(seg(t, 6.55, 7)), ax = lerp(180, 520, ak), waiting = t >= ENT[1] + 0.15 && t < 8.7;+ arrow.setAttribute("d", `M180 221 H${ax.toFixed(1)}`); arrow.setAttribute("opacity", ak > 0.02 ? 1 : 0);+ ahead.setAttribute("transform", `translate(${(ax + 3).toFixed(1)} 221)`); ahead.setAttribute("opacity", ak > 0.06 ? 1 : 0);+ wait.forEach((c, j) => c.setAttribute("opacity", waiting ? (0.25 + 0.75 * Math.max(0, Math.sin(now * 5 - j * 0.9))).toFixed(2) : 0));+ tf(tk[1], 103, 252, land(seg(t, 8.7, 9.1), 2.4).toFixed(3));+ const bi = seg(t, 6.95, 7.6);+ tf(B, (640 + 70 * (1 - cl(bk(bi, 1.3), 0, 1.2)) + 70 * io(out)).toFixed(2), 262, lerp(0.9, 1, oc(bi)).toFixed(3), ((1 - oc(bi)) * 4).toFixed(2));+ B.setAttribute("opacity", (seg(bi, 0, 0.35) * (1 - seg(out, 0.3, 0.9))).toFixed(2));+ tf(sb, 0, 42, ((1 - 0.08 * Math.sin(Math.PI * seg(t, 8.05, 8.35))) * (1 - seg(t, 8.4, 8.55))).toFixed(3));+ tf(bt, 0, 42, (land(seg(t, 8.5, 8.9), 2.6) * (1 + 0.04 * seg(t, 9.6, 10.2) * Math.sin(now * 3))).toFixed(3));+ /* line three: a version number, drawn as blocks */+ ver.forEach((r, j) => tf(r, X0 + 7.5 + 25 * j, 320, land(seg(t, 11 + 0.1 * j, 11.25 + 0.1 * j), 2.2).toFixed(3)));+ vdot.forEach((c, j) => c.setAttribute("opacity", t >= 11.12 + 0.1 * j ? 1 : 0));+ tf(tk[2], 181, 320, land(seg(t, 11.35, 11.75), 2.4).toFixed(3));+ vt.setAttribute("opacity", seg(t, 11.4, 11.6).toFixed(2));+ say(cap, t < TY[0] || out > 0 ? "Open a terminal, paste one block, press Enter" : t < PR[1] ? "Install the engine" : t < PR[2] ? "Run it once to sign in" : t < 11.3 ? "Then ask for its version to check" : "A version number means you are done with the terminal", 700);+ };+ },+ };++ /* ---- download: the download page has one button per system; each lets go of its file, which comes down on a parachute ---- */+ SC.download = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Get the file for your system", "tap a button");+ const SYS = ["macOS", "Windows", "Linux"], COL = [SKY, LIL, MINT], BX = [132, 394, 668], FX = [132, 322, 668], START = [1, 4.4, 8.6], LEN = [2.7, 3.6, 2.7];+ const LINE = ["A universal .dmg for macOS", "A setup .exe or a portable .exe for Windows", "An .AppImage for Linux"];+ const YH = 248, YF = 346, YP = 374, YR = 416, PH = 256, PR = 422, tap = [null, null, null], mid = { "text-anchor": "middle" };+ const bo = (x) => { const n = 7.5625, d = 2.75; return x < 1 / d ? n * x * x : x < 2 / d ? n * (x -= 1.5 / d) * x + 0.75 : x < 2.5 / d ? n * (x -= 2.25 / d) * x + 0.9375 : n * (x -= 2.625 / d) * x + 0.984375; };+ /* the ground, with a landing pad for each system */+ el("path", { d: "M44 490 H756", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ [[46, 172], [238, 312], [582, 172]].forEach(([x, w], i) => el("rect", { x, y: 472, width: w, height: 18, rx: 6, fill: COL[i], stroke: INK, "stroke-width": 3 }, svg));+ const file = (p, w, h, c, a, b) => {+ const g = el("g", {}, p), f = 18;+ el("path", { d: `M${-w} ${-h} H${w - f} L${w} ${-h + f} V${h} H${-w} Z`, fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, g);+ el("path", { d: `M${w - f} ${-h} V${-h + f} H${w} Z`, fill: c, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, g);+ if (a) txt(g, 0, -h + 42, a, Object.assign({ "font-size": 19 }, mid));+ fit(txt(g, 0, a ? -h + 74 : -h + 63, b, Object.assign({ "font-family": MONO, "font-size": a ? 30 : 23, "font-weight": 700 }, mid)), 2 * w - 22);+ el("path", { d: h > 52 ? `M${-w + 18} ${h - 20} H${w - 18} M${-w + 18} ${h - 9} H${w - 52}` : `M${-w + 18} ${h - 10} H${w - 18}`, stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ };+ const guide = [132, 322, 478, 668].map((x) => {+ const g = el("g", {}, svg);+ el("path", { d: `M${x} 322 V446`, stroke: LIL, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, g);+ el("path", { d: `M${x - 10} 446 l10 10 l10 -10`, fill: "none", stroke: LIL, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ });+ const rigs = [["universal", ".dmg"], ["setup", ".exe"], [null, ".AppImage"]].map(([a, b], i) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ const str = el("path", { fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g);+ const can = el("g", {}, g);+ el("path", { d: "M-75 0 C-75 -36 -42 -54 0 -54 C42 -54 75 -36 75 0 Q56 -11 37.5 0 Q19 -11 0 0 Q-19 -11 -37.5 0 Q-56 -11 -75 0 Z", fill: COL[i], stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, can);+ el("path", { d: "M0 -54 Q-24 -32 -37.5 0 M0 -54 V0 M0 -54 Q24 -32 37.5 0", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, can);+ return { g, str, can, f: file(g, 75, 56, COL[i], a, b) };+ });+ /* the portable one has a carry handle, and no parachute */+ const port = mv(el("g", { "pointer-events": "none" }, svg)), HANDLE = "M-16 -50 V-58 a7 7 0 0 1 7 -7 H9 a7 7 0 0 1 7 7 V-50";+ el("path", { d: HANDLE, fill: "none", stroke: INK, "stroke-width": 10, "stroke-linecap": "round" }, port);+ el("path", { d: HANDLE, fill: "none", stroke: YEL, "stroke-width": 4.5, "stroke-linecap": "round" }, port);+ file(port, 59, 50, LIL, "portable", ".exe");+ /* the download page */+ el("rect", { x: 44, y: 92, width: 712, height: 94, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M44 124 H756", stroke: INK, "stroke-width": 3 }, svg);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 262, y: 96, width: 276, height: 24, rx: 12, fill: SOFT }, svg);+ txt(svg, 400, 114, "harnessmd.com/download", Object.assign({ "font-family": MONO, "font-size": 17, "font-weight": 700 }, mid));+ [132, 322, 478, 668].forEach((x) => el("rect", { x: x - 7, y: 182, width: 14, height: 20, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg));+ const btn = SYS.map((s, i) => {+ const g = el("g", {}, svg), x = BX[i];+ el("rect", { x: x - 81, y: 128, width: 162, height: 54, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const f = el("g", {}, g);+ el("rect", { x: x - 76, y: 133, width: 152, height: 42, rx: 12, fill: COL[i], stroke: INK, "stroke-width": 3 }, f);+ el("path", { d: `M${x - 57} 144 V158 M${x - 63} 153 l6 6 l6 -6 M${x - 65} 165 H${x - 49}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, f);+ txt(f, x + 13, 162, s, Object.assign({ "font-size": 23, "font-weight": 700 }, mid));+ button(g, `Drop the ${s} file again`, () => { tap[i] = api.now(); api.poke(); });+ return f;+ });+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const wrap = t >= 13.2;+ if (wrap) tap.fill(null);+ let hot = -1;+ rigs.forEach((R, i) => {+ const lc = t - START[i];+ if (tap[i] !== null && now - tap[i] >= LEN[i] && lc >= LEN[i]) tap[i] = null;+ const c = tap[i] !== null ? now - tap[i] : wrap ? (t < 13.55 ? 99 : -1) : lc;+ if (tap[i] !== null && c < 3.6) hot = i;+ const a = (wrap ? (t < 13.55 ? 1 - seg(t, 13.2, 13.5) : seg(t, 13.6, 13.95)) : 1) * (tap[i] !== null ? seg(c, 0, 0.18) : 1);+ const hop = wrap && t >= 13.55 ? -8 * (1 - oc(seg(t, 13.6, 13.95))) : 0;+ /* let go, fall, open, float, land */+ let y = YH, s = 0, sw = 0, q = 0;+ if (c > 0.2 && c < 0.65) { const k = (c - 0.2) / 0.45; y = lerp(YH, YF, k * k); }+ else if (c >= 0.65) {+ s = land(seg(c, 0.65, 1), 2);+ y = c < 0.95 ? lerp(YF, YP, oc((c - 0.65) / 0.3)) : lerp(YP, YR, io(seg(c, 0.95, 2.2)));+ sw = Math.sin((c - 0.65) * 4.4) * (1 - seg(c, 1.4, 2.2));+ if (c > 2.2) q = Math.exp(-7 * (c - 2.2)) * Math.cos(18 * (c - 2.2));+ }+ const idle = seg(c, 2.4, 3.2), dx = 2.5 * idle * Math.sin(now * 1.3 + i * 2), sy = s * (1 + 0.04 * idle * Math.sin(now * 2.2 + i * 1.7));+ R.g.setAttribute("transform", `translate(${(FX[i] + 7 * sw).toFixed(2)} ${(y + hop).toFixed(2)}) rotate(${(5 * sw).toFixed(2)})`);+ R.g.setAttribute("opacity", a.toFixed(2));+ R.f.setAttribute("transform", `translate(0 56) scale(${(1 + 0.04 * q).toFixed(4)} ${(1 - 0.06 * q).toFixed(4)}) translate(0 -56)`);+ if (s < 0.02) { R.can.setAttribute("display", "none"); R.str.setAttribute("display", "none"); }+ else {+ R.can.removeAttribute("display"); R.str.removeAttribute("display");+ R.can.setAttribute("transform", `translate(${dx.toFixed(2)} -106) scale(${s.toFixed(4)} ${sy.toFixed(4)})`);+ R.str.setAttribute("d", [[-75, -46], [-37.5, -46], [37.5, 46], [75, 46]].map(([u, v]) => `M${(u * s + dx).toFixed(1)} -106 L${v} -56`).join(" "));+ }+ btn[i].setAttribute("transform", `translate(0 ${(4.5 * Math.sin(Math.PI * seg(c, 0, 0.3))).toFixed(2)})`);+ const gv = wrap ? (t < 13.55 ? 0 : seg(t, 13.6, 13.95)) : 1, fade = (j, cc) => guide[j].setAttribute("opacity", (gv * (1 - seg(cc, 0.05, 0.4))).toFixed(2));+ if (i === 1) { fade(1, c); fade(2, c - 1.3); } else fade(i ? 3 : 0, c);+ if (i === 1) {+ const k = seg(c - 1.3, 0, 1);+ tf(port, 478, (lerp(PH, PR, bo(k)) + hop).toFixed(2), 1, (k > 0 && k < 1 ? 9 * Math.sin(k * 10) * (1 - k) : 0).toFixed(2));+ port.setAttribute("opacity", a.toFixed(2));+ }+ });+ say(cap, hot >= 0 ? LINE[hot] : t < START[0] || wrap ? "Get the file for your system from the download page" : t < START[1] ? LINE[0] : t < START[2] ? LINE[1] : t < 11.4 ? LINE[2] : "One file per system", 700);+ };+ },+ };++ /* ---- signin: two windows; the browser shows a code and a copy of it is carried across to the app ---- */+ SC.signin = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Sign in to continue", "tap the app window");+ const AX = 222, FY = 306, mid = { "text-anchor": "middle" };+ const tickDot = (p, r) => {+ const g = el("g", {}, p);+ el("circle", { r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: `M${(-0.46 * r).toFixed(1)} ${(0.04 * r).toFixed(1)} l${(0.34 * r).toFixed(1)} ${(0.36 * r).toFixed(1)} l${(0.6 * r).toFixed(1)} ${(-0.76 * r).toFixed(1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* the code is never letters: six boxes, a dot in each */+ const chip = (p) => {+ const g = el("g", {}, p), d = [];+ for (let j = 0; j < 6; j++) {+ const b = el("g", { transform: `translate(${-95 + 38 * j} 0)` }, g);+ el("rect", { x: -16, y: -20, width: 32, height: 40, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("circle", { r: 5, fill: INK }, b);+ d.push(b);+ }+ return { g, d };+ };+ const ring = (p, w, h) => { const r = el("rect", { fill: "none", stroke: BLUE, "stroke-width": 3, opacity: 0 }, p); r._w = w; r._h = h; return r; };+ const pulse = (r, x, y, k) => {+ const d = 13 * oc(k);+ r.setAttribute("x", (x - r._w / 2 - d).toFixed(1)); r.setAttribute("y", (y - r._h / 2 - d).toFixed(1));+ r.setAttribute("width", (r._w + 2 * d).toFixed(1)); r.setAttribute("height", (r._h + 2 * d).toFixed(1)); r.setAttribute("rx", (r._h / 2 + d).toFixed(1));+ r.setAttribute("opacity", k > 0 && k < 1 ? (0.85 * (1 - k)).toFixed(2) : 0);+ };+ /* where the browser will be: it is not open yet */+ const ph = el("g", {}, svg);+ el("rect", { x: 420, y: 128, width: 336, height: 328, rx: 14, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, ph);+ el("circle", { cx: 588, cy: 292, r: 48, fill: LIL, "fill-opacity": 0.3, stroke: LIL, "stroke-width": 3 }, ph);+ el("path", { d: "M540 292 H636 M588 244 C560 268 560 316 588 340 M588 244 C616 268 616 316 588 340", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, ph);+ el("path", { d: "M420 164 H756", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-dasharray": "1 10" }, ph);+ el("rect", { x: 498, y: 376, width: 180, height: 16, rx: 8, fill: LIL, "fill-opacity": 0.3 }, ph);+ el("rect", { x: 528, y: 404, width: 120, height: 16, rx: 8, fill: LIL, "fill-opacity": 0.3 }, ph);+ /* the app window: step one of the four */+ const app = el("g", {}, svg);+ el("rect", { x: 37, y: 85, width: 370, height: 412, rx: 20, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, app);+ el("rect", { x: 44, y: 92, width: 356, height: 398, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, app);+ el("path", { d: "M44 124 H400", stroke: INK, "stroke-width": 3 }, app);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, app));+ const step = [0, 1, 2, 3].map((j) => el("circle", { cx: 316 + 20 * j, cy: 108, r: 6, fill: j ? "#fff" : YEL, stroke: INK, "stroke-width": 2.5 }, app));+ const r1 = ring(app, 308, 52), b1 = el("g", {}, app);+ el("rect", { x: -154, y: -26, width: 308, height: 52, rx: 26, fill: YEL, stroke: INK, "stroke-width": 3 }, b1);+ fit(txt(b1, 0, 8, "Open browser to sign in", Object.assign({ "font-size": 22, "font-weight": 700 }, mid)), 272);+ const lab = txt(app, 70, 262, "Code from the browser", { "font-size": 20, fill: MUTE });+ el("rect", { x: 68, y: 276, width: 308, height: 60, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, app);+ const ghost = el("g", {}, app);+ for (let j = 0; j < 6; j++) el("rect", { x: AX - 111 + 38 * j, y: FY - 20, width: 32, height: 40, rx: 7, fill: "none", stroke: LIL, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-dasharray": "1 7" }, ghost);+ const r2 = ring(app, 220, 50), b2 = el("g", {}, app);+ const b2r = el("rect", { x: -110, y: -25, width: 220, height: 50, rx: 25, fill: SOFT, stroke: INK, "stroke-width": 3 }, b2);+ const b2t = txt(b2, 0, 8, "Use this code", Object.assign({ "font-size": 22, "font-weight": 700, fill: MUTE }, mid));+ const done = pill(app, "You are signed in", MINT, true);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 15, YEL);+ button(app, "Watch the sign in again", () => api.seek(0.3));+ /* the browser: more tabs than anyone needs, and the code */+ const B = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -168, y: -164, width: 336, height: 328, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, B);+ el("path", { d: "M-168 -128 V-150 a14 14 0 0 1 14 -14 H154 a14 14 0 0 1 14 14 V-128 Z", fill: SOFT, stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, B);+ const tabs = Array.from({ length: 9 }, (_, j) => ({ r: el("rect", { y: -154, height: 26, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2 }, B), c: el("circle", { cy: -141, r: 3.5, fill: [BLUE, MINT, LIL, SKY][j % 4] }, B) }));+ const nt = el("rect", { y: -154, height: 26, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2 }, B);+ el("rect", { x: -150, y: -118, width: 300, height: 22, rx: 11, fill: SOFT }, B);+ const bt = tickDot(B, 17);+ txt(B, 0, -30, "copy the code", Object.assign({ "font-size": 21, fill: MUTE }, mid));+ const sel = el("rect", { x: -121, y: -14, width: 242, height: 56, rx: 10, fill: SKY, opacity: 0 }, B);+ const code = chip(B); code.g.setAttribute("transform", "translate(0 14)");+ el("rect", { x: -100, y: 72, width: 200, height: 14, rx: 7, fill: SOFT }, B);+ el("rect", { x: -70, y: 98, width: 140, height: 14, rx: 7, fill: SOFT }, B);+ const T = chip(svg); mv(T.g); T.g.setAttribute("pointer-events", "none");+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.7), vis = 1 - out, live = t < 13.2;+ /* press Open browser to sign in */+ tf(b1, AX, 176, (1 - 0.05 * Math.sin(Math.PI * seg(t, 1.3, 1.65)) + (t < 1.3 ? 0.015 * Math.sin(now * 4) : 0)).toFixed(4));+ pulse(r1, AX, 176, seg(t, 1.45, 2.05));+ /* the browser opens where the outline was, and one more tab squeezes in */+ const bo = seg(t, 1.8, 2.45), bs = cl(bk(bo, 1.5), 0, 1.2) * (1 - io(out));+ tf(B, 588, 292, bs.toFixed(4), ((1 - oc(bo)) * -4).toFixed(2));+ B.setAttribute("opacity", bs > 0.02 ? 1 : 0);+ ph.setAttribute("opacity", cl(1 - seg(t, 1.8, 2.1) + seg(t, 13.45, 13.7)).toFixed(2));+ const pitch = lerp(35, 25.2, io(seg(t, 2.35, 3.05)));+ tabs.forEach((o, j) => { o.r.setAttribute("x", (-159 + j * pitch).toFixed(1)); o.r.setAttribute("width", (pitch - 3).toFixed(1)); o.c.setAttribute("cx", (-160.5 + (j + 0.5) * pitch).toFixed(1)); });+ nt.setAttribute("x", (-159 + 9 * pitch).toFixed(1)); nt.setAttribute("width", Math.max(0.1, 317 - 9 * pitch).toFixed(1)); nt.setAttribute("opacity", pitch < 34.6 ? 1 : 0);+ tf(bt, 0, -72, land(seg(t, 2.85, 3.25), 2.4).toFixed(3));+ code.d.forEach((b, j) => tf(b, -95 + 38 * j, 0, land(seg(t, 3.25 + 0.1 * j, 3.55 + 0.1 * j), 2.2).toFixed(3)));+ sel.setAttribute("opacity", (0.9 * seg(t, 4.45, 4.75) * (1 - seg(t, 6.5, 6.9))).toFixed(2));+ /* a copy of the code lifts out, goes across and drops into the field */+ const k = seg(t, 5.1, 6.3), up = oc(seg(t, 4.85, 5.1)) * (1 - io(seg(t, 6.15, 6.4))), tau = t - 6.4;+ tf(T.g, lerp(588, AX, io(k)).toFixed(2), (FY - 11 * up).toFixed(2), (1 + 0.07 * up + (tau > 0 ? 0.05 * Math.exp(-7 * tau) * Math.sin(18 * tau) : 0)).toFixed(4));+ T.g.setAttribute("opacity", t >= 4.85 ? vis.toFixed(2) : 0);+ ghost.setAttribute("opacity", cl(1 - seg(t, 6, 6.3) + out).toFixed(2));+ lab.setAttribute("fill", t >= 5.1 && live ? INK : MUTE);+ /* press Use this code */+ const ready = t >= 6.55 && live;+ b2r.setAttribute("fill", ready ? YEL : SOFT); b2t.setAttribute("fill", ready ? INK : MUTE);+ tf(b2, AX, 389, (1 + 0.06 * Math.sin(Math.PI * seg(t, 6.55, 6.9)) - 0.05 * Math.sin(Math.PI * seg(t, 7.7, 8.05))).toFixed(4));+ pulse(r2, AX, 389, seg(t, 7.85, 8.45));+ const dk = seg(t, 8.45, 8.9);+ tf(done, 210, 452, (land(dk, 2.2) * vis).toFixed(3), ((1 - oc(dk)) * -6).toFixed(2));+ step[0].setAttribute("fill", t >= 8.45 && live ? MINT : YEL);+ tf(sp, 364, 452, (land(seg(t, 8.8, 9.25), 2.6) * vis * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, t < 1.3 || out > 0 ? "The browser never opens by itself" : t < 2.9 ? "Press Open browser to sign in" : t < 5.1 ? "Copy the code from the browser" : t < 7.4 ? "Paste it into Code from the browser" : t < 8.45 ? "Press Use this code" : "Sign in takes two windows", 700);+ };+ },+ };++ /* ---- fork: a mini golf lane that splits in three, one branch per button; the ball takes the free one and circles the hole first ---- */+ SC.fork = {+ dur: 14, still: 9.2,+ build(svg, api) {+ head(svg, "Run your agents like a company", "tap a branch");+ const LY = [180, 304, 428], HX = 556, FK = 178, A0 = -2.2;+ const BTN = ["Continue with the free version", "I have a licence key", "Get PRO"];+ const DEST = [["Classic view"], ["paste a key"], ["checkout in", "your browser"]];+ const LINE = ["Continue with the free version opens the Classic view", "I have a licence key lets you paste a key", "Get PRO opens the checkout in your browser"];+ const lane = (i) => `M190 304 C250 304 250 ${LY[i]} 310 ${LY[i]} H${HX}`;+ let pick = 0, lastT = 0, held = false;+ /* rough grass, a bunker and a star around the course */+ [[138, 178], [166, 246], [236, 474], [694, 362], [472, 494], [724, 246]].forEach(([x, y]) => el("path", { d: `M${x - 7} ${y} l-3 -11 M${x} ${y} v-14 M${x + 7} ${y} l3 -11`, fill: "none", stroke: MINT, "stroke-width": 3.5, "stroke-linecap": "round" }, svg));+ el("path", { d: "M50 454 C40 422 82 406 112 416 C150 404 176 432 160 460 C148 488 68 488 50 454 Z", fill: YEL, stroke: INK, "stroke-width": 3 }, svg);+ [[84, 440], [112, 456], [136, 438], [96, 468], [140, 464]].forEach(([x, y]) => el("circle", { cx: x, cy: y, r: 2, fill: INK }, svg));+ /* the felt: every lane is drawn twice, ink under mint, so the edge stays one even line */+ const felt = "M64 304 H190 " + [0, 1, 2].map(lane).join(" ");+ el("path", { d: felt, fill: "none", stroke: INK, "stroke-width": 56, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ el("path", { d: felt, fill: "none", stroke: MINT, "stroke-width": 50, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ LY.forEach((y) => el("circle", { cx: HX, cy: y, r: 13, fill: INK }, svg));+ el("rect", { x: 78, y: 287, width: 38, height: 34, rx: 6, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 44, 366, "after setup", { "font-size": 20, fill: MUTE });+ /* one button over each branch; the branch and its button are one tap target */+ const B = BTN.map((s, i) => {+ const g = el("g", {}, svg);+ el("path", { d: lane(i), fill: "none", stroke: "#fff", "stroke-opacity": 0, "stroke-width": 56, "stroke-linecap": "round" }, g);+ const f = el("g", {}, g), t = fit(txt(f, 0, 7, s, { "text-anchor": "middle", "font-size": 19, "font-weight": 700 }), 286);+ const w = Math.round(t.getComputedTextLength() + 38);+ const r = el("rect", { x: -w / 2, y: -20, width: w, height: 40, rx: 20, fill: "#fff", stroke: INK, "stroke-width": 3 });+ f.insertBefore(r, t);+ f.insertBefore(el("rect", { x: -w / 2 - 5, y: -25, width: w + 10, height: 50, rx: 25, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }), r);+ const pr = el("rect", { fill: "none", stroke: BLUE, "stroke-width": 3, opacity: 0 }, g);+ button(g, `Take the branch ${s}`, () => { pick = i; held = true; api.seek(3.4); });+ return { f, r, pr, w, cx: 318 + w / 2, y: LY[i] - 42 };+ });+ const D = DEST.map((L, i) => L.map((s, j) => txt(svg, 604, LY[i] + (L.length > 1 ? -5 + 24 * j : 7), s, { "font-size": 20, "font-weight": 700, fill: MUTE })));+ const disc = [0, 1, 2].map((i) => {+ const g = el("g", { "pointer-events": "none" }, svg);+ el("circle", { r: 16, fill: i ? LIL : MINT, stroke: INK, "stroke-width": 2.5 }, g);+ if (i) el("circle", { r: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ else el("path", { d: "M-7 1 l5 5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ });+ /* the putter, and the ball */+ const put = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M0 0 V190", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, put);+ el("rect", { x: -6, y: -4, width: 12, height: 56, rx: 6, fill: LIL, stroke: INK, "stroke-width": 2.5 }, put);+ el("rect", { x: -10, y: 186, width: 24, height: 27, rx: 5, fill: YEL, stroke: INK, "stroke-width": 2.5 }, put);+ const ball = mv(el("g", { "pointer-events": "none" }, svg));+ el("circle", { r: 11, fill: "#fff", stroke: INK, "stroke-width": 3 }, ball);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 13, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ if (t < lastT - 1 && !held) pick = 0;+ held = false; lastT = t;+ const live = t < 13.2, out = seg(t, 13.2, 13.5), P = pick, Y = LY[P];+ /* the putt */+ const th = t < 0.75 ? 5 * io(seg(t, 0.3, 0.75)) : t < 0.9 ? lerp(5, -2, Math.pow(seg(t, 0.75, 0.9), 2)) : -2 * (1 - oc(seg(t, 0.9, 1.5)));+ tf(put, 62, 104, 1, th.toFixed(2));+ let x = 96, y = 304 + (t < 0.3 ? 0.6 * Math.sin(now * 5) : 0), s = 1, o = 1;+ if (live && t >= 0.88) {+ /* it rolls to the split, stops, and leans toward two branches before it picks */+ const h2 = Math.sin(Math.PI * seg(t, 2.1, 2.7)), h1 = Math.sin(Math.PI * seg(t, 2.8, 3.35));+ x = lerp(96, FK, oc(seg(t, 0.88, 1.9))) + 9 * h2 + 15 * h1; y = 304 + 8 * h2;+ const g = seg(t, 4, 5.9);+ if (g > 0) {+ const BL = P === 1 ? 120 : 181, d = (1 - Math.pow(1 - io(g), 1.6)) * (12 + BL + 237);+ if (d <= 12) x = FK + d;+ else if (d <= 12 + BL) { const q = (d - 12) / BL; x = 190 + 180 * q - 180 * q * q + 120 * q * q * q; y = 304 + (Y - 304) * (3 * q * q - 2 * q * q * q); }+ else { const q = (d - 12 - BL) / 237; x = lerp(310, HX + 16 * Math.cos(A0), q); y = Y + 16 * Math.sin(A0) * io(seg(q, 0.55, 1)); }+ }+ /* round the rim once, then in */+ const k = seg(t, 5.9, 7);+ if (k > 0) { const a = A0 + 2.6 * Math.PI * (1 - Math.pow(1 - k, 1.8)), r = 16 * (1 - Math.pow(k, 2.5)); x = HX + r * Math.cos(a); y = Y + r * Math.sin(a); s = 1 - 0.4 * Math.pow(k, 3); o = 1 - seg(k, 0.82, 1); }+ }+ if (!live) o = seg(t, 13.5, 13.9);+ tf(ball, x.toFixed(2), y.toFixed(2), s.toFixed(3)); ball.setAttribute("opacity", o.toFixed(2));+ let say2 = 0;+ B.forEach((b, i) => {+ const a = 9.9 + 1.6 * (i - 1), pv = live && P === 0 && i > 0 && t >= a && t < a + 1.5;+ if (pv) say2 = i;+ const on = (live && i === P && t >= 3.5) || pv;+ b.r.setAttribute("fill", on ? (i ? LIL : YEL) : "#fff");+ tf(b.f, b.cx, b.y, (1 + 0.05 * Math.sin(Math.PI * seg(t, 0.1 + 0.14 * i, 0.5 + 0.14 * i)) - (i === P ? 0.05 * Math.sin(Math.PI * seg(t, 3.5, 3.85)) : 0) + (P === 0 && i > 0 ? 0.05 * Math.sin(Math.PI * seg(t, a, a + 0.4)) : 0)).toFixed(4));+ const k = i === P ? seg(t, 3.65, 4.25) : 0, e = 13 * oc(k);+ b.pr.setAttribute("x", (b.cx - b.w / 2 - e).toFixed(1)); b.pr.setAttribute("y", (b.y - 20 - e).toFixed(1));+ b.pr.setAttribute("width", (b.w + 2 * e).toFixed(1)); b.pr.setAttribute("height", (40 + 2 * e).toFixed(1)); b.pr.setAttribute("rx", (20 + e).toFixed(1));+ b.pr.setAttribute("opacity", k > 0 && k < 1 ? (0.85 * (1 - k)).toFixed(2) : 0);+ D[i].forEach((n) => n.setAttribute("fill", (live && i === P && t >= 7) || pv ? INK : MUTE));+ tf(disc[i], HX, LY[i], (i === P ? land(seg(t, 6.95, 7.35), 2.2) * (1 - out) : 0).toFixed(3));+ });+ tf(sp, 748, 150, (P === 0 ? land(seg(t, 7.3, 7.75), 2.6) * (1 - out) * (1 + 0.12 * Math.sin(now * 3)) : 0).toFixed(3), (now * 20).toFixed(1));+ say(cap, !live || t < 2 ? "The first screen after setup" : t < 3.5 ? "The screen that stops people" : say2 ? LINE[say2] : P === 0 && t < 7 ? "Without a licence, press Continue with the free version" : LINE[P], 700);+ };+ },+ };++ /* ---- ide: the editor window; an agent adds two lines, Changes marks them, and the switch shows the file as code, split or preview ---- */+ SC.ide = {+ dur: 14, still: 11.6,+ build(svg, api) {+ head(svg, "What is the IDE button for?", "tap the switch");+ const mid = { "text-anchor": "middle" }, PX = 224, PR = 756, PW = PR - PX, TOP = 169.5, BOT = 488.5;+ const LN = (j) => 194 + 25 * j, FY = (j) => 196 + 37 * j, X0 = PX + 48, TW = [360, 236];+ let uv = -1, ut = 0, v0 = 0, last = 0;+ /* the window: title bar, tool row, file list at the side */+ el("rect", { x: 44, y: 92, width: 712, height: 398, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: `M44 124 H756 M44 168 H756 M${PX} 168 V490`, stroke: INK, "stroke-width": 3 }, svg);+ [LIL, YEL, MINT].forEach((c, i) => el("circle", { cx: 68 + 19 * i, cy: 108, r: 5.5, fill: c, stroke: INK, "stroke-width": 2.5 }, svg));+ el("rect", { x: 132, y: 97, width: 54, height: 22, rx: 11, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 159, 113.5, "IDE", Object.assign({ "font-size": 16, "font-weight": 700 }, mid));+ el("rect", { x: 262, y: 96, width: 276, height: 24, rx: 12, fill: SOFT }, svg);+ fit(txt(svg, 400, 114, "an agent's working folder", Object.assign({ "font-size": 17 }, mid)), 250);+ let tx = 58;+ const tabs = ["Changes", "History", "Compare"].map((s) => {+ const g = el("g", {}, svg), t = txt(g, 0, 152.5, s, Object.assign({ "font-size": 18, "font-weight": 700 }, mid)), w = Math.round(t.getComputedTextLength() + 26);+ const r = el("rect", { x: tx, y: 131, width: w, height: 30, rx: 9, fill: "#fff", stroke: INK, "stroke-width": 2.5 }), o = { g, r, cx: tx + w / 2 };+ g.insertBefore(r, t); t.setAttribute("x", o.cx); tx += w + 8;+ return o;+ });+ /* the switch: three words, and a yellow knob under the one in use */+ const SW = ["code", "split", "preview"], swg = el("g", {}, svg);+ const swbg = el("rect", { y: 131, height: 30, rx: 15, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, swg), knob = el("rect", { y: 131, height: 30, rx: 15, fill: YEL, stroke: INK, "stroke-width": 2.5 }, swg);+ let sx = 0;+ const S3 = SW.map((s) => { const g = el("g", {}, swg), t = txt(g, 0, 152.5, s, Object.assign({ "font-size": 18, "font-weight": 700 }, mid)), w = Math.round(t.getComputedTextLength() + 28), o = { g, t, w, x: sx }; sx += w; return o; });+ swbg.setAttribute("x", 744 - sx); swbg.setAttribute("width", sx);+ S3.forEach((o, i) => {+ o.x += 744 - sx; o.t.setAttribute("x", o.x + o.w / 2);+ o.g.insertBefore(el("rect", { x: o.x - 4, y: 127, width: o.w + 8, height: 38, rx: 19, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }), o.t);+ button(o.g, `Show the file as ${SW[i]}`, () => { v0 = last; uv = i; ut = api.now(); api.poke(); });+ });+ el("rect", { x: 52, y: FY(2) - 16, width: 164, height: 32, rx: 8, fill: SOFT }, svg);+ [84, 62, 96, 54, 78, 70].forEach((w, j) => {+ const g = el("g", { transform: `translate(${j ? 84 : 72} ${FY(j)})` }, svg);+ if (!j) el("path", { d: "M-9 -7 H-2 L1 -4 H9 V7 H-9 Z", fill: YEL, stroke: INK, "stroke-width": 2.2, "stroke-linejoin": "round" }, g);+ else if (j === 4) { el("rect", { x: -8, y: -7, width: 16, height: 14, rx: 2, fill: SKY, stroke: INK, "stroke-width": 2.2 }, g); el("path", { d: "M-6 5 l5 -6 l4 4 l2 -2 l3 4", fill: "none", stroke: INK, "stroke-width": 1.8, "stroke-linejoin": "round" }, g); }+ else el("path", { d: "M-7 -9 H2 L7 -4 V9 H-7 Z", fill: "#fff", stroke: INK, "stroke-width": 2.2, "stroke-linejoin": "round" }, g);+ el("rect", { x: 18, y: -5, width: w, height: 10, rx: 5, fill: MUTE, "fill-opacity": j === 2 ? 0.85 : 0.4 }, g);+ });+ const dot = el("circle", { r: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, svg);+ el("rect", { x: 62, y: 438, width: 144, height: 36, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M82 456 h14 M89 449 v14", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ txt(svg, 106, 462.5, "New file", { "font-size": 18, "font-weight": 700 });+ /* the pane: code on the left of the divider, the formatted preview on the right of it */+ const defs = el("defs", {}, svg);+ const cc = el("rect", { x: PX + 1.5, y: TOP, height: BOT - TOP }, el("clipPath", { id: "mg-ide-code" }, defs)), cp = el("rect", { y: TOP, height: BOT - TOP }, el("clipPath", { id: "mg-ide-prev" }, defs));+ const C = el("g", { "clip-path": "url(#mg-ide-code)" }, svg), V = el("g", { "clip-path": "url(#mg-ide-prev)" }, svg);+ el("rect", { x: PX + 1.5, y: TOP, width: 32, height: BOT - TOP, fill: SOFT }, C);+ const band = [9, 10].map((j) => el("rect", { x: PX + 33.5, y: LN(j) - 12.5, width: PW, height: 25, fill: MINT, "fill-opacity": 0.4, opacity: 0 }, C));+ const mark = [9, 10].map((j) => el("rect", { x: PX + 4, y: LN(j) - 10, width: 6, height: 20, rx: 3, fill: MINT, stroke: INK, "stroke-width": 2, opacity: 0 }, C));+ for (let j = 0; j < 11; j++) el("rect", { x: PX + 15, y: LN(j) - 2, width: 12, height: 4, rx: 2, fill: MUTE, "fill-opacity": 0.45 }, C);+ const bar = (x, j, w, fill, op, h = 10) => el("rect", { x, y: LN(j) - h / 2, width: w, height: h, rx: h / 2, fill, "fill-opacity": op }, C);+ el("rect", { x: X0, y: LN(0) - 6, width: 12, height: 12, rx: 3, fill: LIL }, C); bar(X0 + 20, 0, 150, INK, 0.9, 12);+ bar(X0, 2, 400, MUTE, 0.5); bar(X0, 3, 330, MUTE, 0.5);+ [190, 250, 160].forEach((w, i) => { el("rect", { x: X0, y: LN(5 + i) - 2, width: 9, height: 4, rx: 2, fill: BLUE }, C); bar(X0 + 18, 5 + i, w, MUTE, 0.5); });+ const typed = [bar(X0, 9, 0, INK, 0.75), bar(X0, 10, 0, INK, 0.75)];+ const cur = el("rect", { width: 3, height: 20, fill: INK }, C);+ const pvH = el("rect", { y: 184, height: 20, rx: 10, fill: INK }, V);+ const pvP = [232, 254].map((y) => el("rect", { y, height: 10, rx: 5, fill: MUTE, "fill-opacity": 0.5 }, V));+ const pvL = [298, 324, 350].map((y) => ({ d: el("circle", { cy: y + 5, r: 5, fill: BLUE }, V), b: el("rect", { y, height: 10, rx: 5, fill: MUTE, "fill-opacity": 0.5 }, V) }));+ const pvN = [398, 420].map((y) => el("rect", { y, height: 10, rx: 5, fill: INK, "fill-opacity": 0.75 }, V));+ const div = el("path", { stroke: INK, "stroke-width": 3 }, svg);+ /* the scroll bar: it is a long report */+ el("rect", { x: 740, y: 178, width: 8, height: 302, rx: 4, fill: LIL, "fill-opacity": 0.3 }, svg);+ const thumb = el("rect", { x: 740, y: 181, width: 8, rx: 4, fill: MUTE }, svg);+ const cap = strip(svg, 508, 740);+ const auto = (t) => (t < 7.6 ? io(seg(t, 5.2, 5.85)) : t < 9.8 ? 1 + io(seg(t, 7.6, 8.25)) : t < 13.2 ? 2 - io(seg(t, 9.8, 10.45)) : 1 - io(seg(t, 13.2, 13.7)));+ return (t, now) => {+ const live = t < 13.2, out = seg(t, 13.2, 13.5);+ let v = auto(t);+ if (uv >= 0) { const c = now - ut; if (c >= 5.6) uv = -1; else v = c < 5 ? lerp(v0, uv, io(seg(c, 0, 0.45))) : lerp(uv, v, io(seg(c, 5, 5.6))); }+ last = v;+ const dx = PR - (v / 2) * PW, x = dx + 24, av = Math.max(0, PR - dx - 58);+ cc.setAttribute("width", Math.max(0, dx - PX - 1.5).toFixed(1));+ cp.setAttribute("x", dx.toFixed(1)); cp.setAttribute("width", Math.max(0, PR - 1.5 - dx).toFixed(1));+ div.setAttribute("d", `M${dx.toFixed(1)} 168 V${(dx > PR - 14 ? 476 + Math.sqrt(196 - Math.pow(dx - PR + 14, 2)) : 490).toFixed(1)}`);+ const a = Math.min(1, Math.floor(v)), f = v - a;+ knob.setAttribute("x", lerp(S3[a].x, S3[a + 1].x, f).toFixed(1)); knob.setAttribute("width", lerp(S3[a].w, S3[a + 1].w, f).toFixed(1));+ /* the agent types two more lines, and the file in the list gets its dot */+ const ty = [seg(t, 0.9, 1.7), seg(t, 1.8, 2.4)].map((k) => (Math.round(k * 12) / 12) * (1 - out)), ci = t >= 1.8 ? 1 : 0;+ typed.forEach((n, i) => n.setAttribute("width", (TW[i] * ty[i]).toFixed(1)));+ cur.setAttribute("x", (X0 + TW[ci] * ty[ci] + 4).toFixed(1)); cur.setAttribute("y", LN(9 + ci) - 10);+ cur.setAttribute("opacity", Math.sin(now * 6) > -0.2 ? 1 : 0);+ tf(dot, 204, FY(2), (land(seg(t, 0.95, 1.35), 2.4) * (1 - out)).toFixed(3));+ thumb.setAttribute("height", (15 - 2 * (ty[0] + ty[1])).toFixed(1));+ /* Changes lights up and marks the two new lines */+ tabs[0].r.setAttribute("fill", live && t >= 3.2 ? YEL : "#fff");+ tabs[0].g.setAttribute("transform", `translate(${tabs[0].cx} 146) scale(${(1 + 0.09 * Math.sin(Math.PI * seg(t, 3.2, 3.6))).toFixed(4)}) translate(${-tabs[0].cx} -146)`);+ band.forEach((n, i) => { const k = (seg(t, 3.5 + 0.2 * i, 3.8 + 0.2 * i) * (1 - out)).toFixed(2); n.setAttribute("opacity", k); mark[i].setAttribute("opacity", k); });+ /* the preview keeps its own layout, whatever room it has */+ pvH.setAttribute("x", x.toFixed(1)); pvH.setAttribute("width", (0.62 * av).toFixed(1));+ pvP.forEach((n, i) => { n.setAttribute("x", x.toFixed(1)); n.setAttribute("width", ([1, 0.78][i] * av).toFixed(1)); });+ pvL.forEach((o, i) => { o.d.setAttribute("cx", (x + 6).toFixed(1)); o.b.setAttribute("x", (x + 22).toFixed(1)); o.b.setAttribute("width", ([0.55, 0.72, 0.45][i] * Math.max(0, av - 22)).toFixed(1)); });+ pvN.forEach((n, i) => { n.setAttribute("x", x.toFixed(1)); n.setAttribute("width", ([0.9, 0.58][i] * av * ty[i]).toFixed(1)); });+ say(cap, uv >= 0 || (t >= 5.2 && t < 9.8) ? "Preview Markdown as code, split or preview" : !live || t < 0.9 ? "The IDE button opens a code editor on an agent's working folder" : t < 3.2 ? "See what an agent just did without leaving the office" : t < 5.2 ? "See what the agent changed: Changes, History and Compare" : "Read the report in preview next to the files it describes", 700);+ };+ },+ };++ /* ---- fixes: four stepping stones over a stream, one per fix, each a bigger restart than the last; the pebble stops on the first that works ---- */+ SC.fixes = {+ dur: 14, still: 11.6,+ build(svg, api) {+ head(svg, "What do you do when an install fails?", "tap a stone");+ const mid = { "text-anchor": "middle" }, T0 = 0.9, GAP = 2, B0 = [68, 356];+ const ST = [[176, 344, 56, 36], [312, 326, 60, 39], [456, 308, 64, 42], [608, 290, 68, 45]];+ const NAME = [["Check again"], ["restart", "the agent"], ["reopen", "the app"], ["restart the", "computer"]];+ const STEP = ["Press Check again or Try again", "Close that agent and start it again", "Quit Munder Difflin completely and open it again", "Restart the computer"];+ const LP = ST.map(([x, y, rx]) => [x - rx + 22, y + 2]);+ const tick = (p) => el("path", { d: "M-6 1 l4.5 4.5 l8 -10", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, p);+ let goal = 2, lastT = 0, held = false;+ /* the stream: ripples, lily pads, and a bank on each side */+ const rip = [[150, 250], [268, 244], [392, 232], [512, 224], [150, 446], [304, 458], [566, 442], [644, 474], [440, 482]].map(([x, y]) => ({ x, y, n: el("path", { d: "M-16 0 q8 -7 16 0 t16 0", fill: "none", stroke: SKY, "stroke-width": 3.5, "stroke-linecap": "round" }, svg) }));+ const pad = [[230, 470, 15], [376, 448, 13], [506, 458, 17], [708, 462, 14]].map(([x, y, r]) => ({ x, y, n: el("path", { d: `M0 0 L${r} 0 A${r} ${r} 0 1 1 ${(r * 0.8).toFixed(1)} ${(-r * 0.6).toFixed(1)} Z`, fill: MINT, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg) }));+ el("path", { d: "M36 336 C32 304 76 290 94 310 C112 330 108 392 88 408 C68 424 42 412 36 386 Z", fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ el("path", { d: "M696 300 C702 266 750 262 764 298 C772 330 768 390 748 406 C726 420 700 398 694 364 Z", fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ [[54, 330], [84, 394], [722, 302], [746, 378], [718, 352]].forEach(([x, y]) => el("path", { d: `M${x - 5} ${y} l-2 -8 M${x} ${y} v-10 M${x + 5} ${y} l2 -8`, fill: "none", stroke: INK, "stroke-width": 2.2, "stroke-linecap": "round" }, svg));+ /* the stones: the same restart arrow on each, a size bigger every time, around the thing it restarts */+ const S = ST.map(([x, y, rx, ry], k) => {+ const g = el("g", {}, svg), R = [18, 20, 22, 25][k];+ el("ellipse", { cx: x, cy: y, rx: rx + 10, ry: ry + 8, fill: "none", stroke: SKY, "stroke-width": 3 }, g);+ const rp = el("ellipse", { cx: x, cy: y, fill: "none", stroke: SKY, "stroke-width": 3, opacity: 0 }, g);+ el("ellipse", { cx: x, cy: y, rx: rx + 5, ry: ry + 5, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const e = el("ellipse", { cx: x, cy: y, rx, ry, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ const io2 = el("g", { transform: `translate(${x + 15} ${y})` }, g), sp = el("g", {}, io2), ic = el("g", { transform: `scale(${1.15 + 0.05 * k})` }, io2), ex = (R * 0.5).toFixed(1), ey = (-R * 0.866).toFixed(1);+ el("path", { d: `M${R} 0 A${R} ${R} 0 1 1 ${ex} ${ey} M${ex} ${ey} l-3 -6.3 M${ex} ${ey} l-7 0.6`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, sp);+ if (k === 0) el("rect", { x: -9, y: -5, width: 18, height: 10, rx: 5, fill: YEL, stroke: INK, "stroke-width": 2.2 }, ic);+ if (k === 1) { el("rect", { x: -10, y: -8, width: 20, height: 16, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M-5 -3 l4 3 l-4 3 M1 4 h5", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round", "stroke-linejoin": "round" }, ic); }+ if (k === 2) { el("rect", { x: -12, y: -9, width: 24, height: 18, rx: 3, fill: LIL, stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M-12 -3 H12", stroke: INK, "stroke-width": 2.2 }, ic); }+ if (k === 3) { el("rect", { x: -14, y: -12, width: 28, height: 19, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2.2 }, ic); el("path", { d: "M0 7 V12 M-8 12.5 H8", stroke: INK, "stroke-width": 2.2, "stroke-linecap": "round" }, ic); }+ NAME[k].forEach((s, j) => txt(svg, x, y + ry + 28 + 23 * j, s, Object.assign({ "font-size": 20, "font-weight": 700 }, mid)));+ button(g, `Hop as far as ${NAME[k].join(" ")}`, () => { goal = k; held = true; api.seek(0.7); });+ return { e, sp, rp, rx, ry, bx: x + 0.707 * rx, by: y - 0.707 * ry };+ });+ S.forEach((s) => {+ s.bd = el("g", { "pointer-events": "none" }, svg);+ s.no = el("g", {}, s.bd); el("circle", { r: 11, fill: LIL, stroke: INK, "stroke-width": 2.5 }, s.no); el("path", { d: "M-4 -4 L4 4 M4 -4 L-4 4", stroke: INK, "stroke-width": 2.8, "stroke-linecap": "round" }, s.no);+ s.ok = el("g", {}, s.bd); el("circle", { r: 11, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, s.ok); tick(s.ok).setAttribute("transform", "scale(0.85)");+ });+ /* the install that failed: a crooked card with a broken bar */+ const card = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: -108, y: -56, width: 216, height: 112, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, card);+ txt(card, -88, -14, "Install", { "font-size": 24, "font-weight": 700 });+ el("rect", { x: -88, y: 10, width: 96, height: 18, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, card);+ const brk = el("g", {}, card); el("rect", { x: 0, y: -9, width: 76, height: 18, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, brk);+ const mend = el("rect", { x: -88, y: 10, width: 176, height: 18, rx: 6, fill: MINT, stroke: INK, "stroke-width": 2.5, opacity: 0 }, card);+ const warn = el("g", {}, card); el("circle", { r: 14, fill: YEL, stroke: INK, "stroke-width": 2.5 }, warn); el("path", { d: "M0 -7 V2", stroke: INK, "stroke-width": 3.2, "stroke-linecap": "round" }, warn); el("circle", { cy: 7.5, r: 2, fill: INK }, warn);+ const okm = el("g", {}, card); el("circle", { r: 14, fill: MINT, stroke: INK, "stroke-width": 2.5 }, okm); tick(okm);+ /* past the far bank: the last resort */+ const bub = el("g", { transform: "translate(650 154)", "pointer-events": "none" }, svg);+ el("rect", { x: -114, y: -50, width: 228, height: 100, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 3 }, bub);+ el("path", { d: "M40 48 L58 80 L76 48", fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round", "stroke-linecap": "round" }, bub);+ txt(bub, 0, -12, "Still stuck?", Object.assign({ "font-size": 20, fill: MUTE }, mid));+ fit(txt(bub, 0, 20, "Ask your orchestrator", Object.assign({ "font-size": 20, "font-weight": 700 }, mid)), 204);+ const sh = el("ellipse", { rx: 10, ry: 4, fill: INK, opacity: 0.16, "pointer-events": "none" }, svg);+ const peb = mv(el("g", { "pointer-events": "none" }, svg));+ el("circle", { r: 11, fill: YEL, stroke: INK, "stroke-width": 3 }, peb); el("circle", { cx: -3.5, cy: -3.5, r: 2.5, fill: "#fff" }, peb);+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 12, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ if (t < lastT - 1 && !held) goal = 2;+ held = false; lastT = t;+ const live = t < 13.2, out = seg(t, 13.2, 13.5), Vg = T0 + GAP * goal + 1.35;+ rip.forEach((o, i) => tf(o.n, (o.x + 4 * Math.sin(now * 0.8 + i * 1.7)).toFixed(2), o.y));+ pad.forEach((o, i) => tf(o.n, (o.x + 5 * Math.sin(now * 0.4 + i * 2)).toFixed(2), (o.y + 2 * Math.cos(now * 0.5 + i)).toFixed(2), 1, (20 * i + 8 * Math.sin(now * 0.3 + i)).toFixed(1)));+ /* the pebble: one hop per stone, as far as the stone that works */+ let px = B0[0], py = B0[1], lift = 0, tl = -1, po = 1;+ for (let k = 0; k <= goal; k++) {+ const h = T0 + GAP * k; if (t < h) break;+ const a = k ? LP[k - 1] : B0, q = seg(t, h, h + 0.5), e = 0.5 * q + 0.5 * io(q);+ px = lerp(a[0], LP[k][0], e); py = lerp(a[1], LP[k][1], e); lift = 4 * q * (1 - q); tl = t - h - 0.5;+ }+ if (!live) { po = t < 13.5 ? 1 - out : seg(t, 13.55, 13.9); if (t >= 13.5) { px = B0[0]; py = B0[1]; lift = 0; tl = -1; } }+ const sq = tl > 0 ? 0.2 * Math.exp(-9 * tl) * Math.cos(20 * tl) : 0, bob = tl > 1 ? 1.2 * Math.sin(now * 2.2) : 0, ps = 1 + 0.22 * lift;+ peb.setAttribute("transform", `translate(${px.toFixed(2)} ${(py - 42 * lift + bob + 11 * sq).toFixed(2)}) scale(${(ps * (1 + 0.5 * sq)).toFixed(4)} ${(ps * (1 - sq)).toFixed(4)})`);+ peb.setAttribute("opacity", po.toFixed(2));+ tf(sh, px.toFixed(2), (py + 10).toFixed(2), (1 - 0.45 * lift).toFixed(3)); sh.setAttribute("opacity", (0.16 * po).toFixed(3));+ S.forEach((s, k) => {+ const h = T0 + GAP * k, used = k <= goal, v = h + 1.35, r = used ? seg(t, h + 0.5, h + 1.25) : 0;+ s.e.setAttribute("fill", used && live && t >= v ? (k === goal ? MINT : SOFT) : "#fff");+ s.sp.setAttribute("transform", `rotate(${used ? (360 * io(seg(t, h + 0.6, h + 1.3))).toFixed(1) : 0})`);+ s.rp.setAttribute("rx", (s.rx + 10 + 20 * oc(r)).toFixed(1)); s.rp.setAttribute("ry", (s.ry + 8 + 14 * oc(r)).toFixed(1)); s.rp.setAttribute("opacity", r > 0 && r < 1 ? (0.9 * (1 - r)).toFixed(2) : 0);+ tf(s.bd, s.bx.toFixed(1), s.by.toFixed(1), (used ? land(seg(t, v, v + 0.4), 2.4) * (1 - out) : 0).toFixed(3));+ s.no.setAttribute("opacity", k === goal ? 0 : 1); s.ok.setAttribute("opacity", k === goal ? 1 : 0);+ });+ /* each try lifts the broken half of the bar; only the last one holds */+ let th = -5 + 0.5 * Math.sin(now * 1.6), ph = 10;+ for (let k = 0; k < goal; k++) { const a = seg(t, T0 + GAP * k + 0.6, T0 + GAP * k + 1.55); if (a > 0 && a < 1) { ph = 10 - 7 * Math.sin(Math.PI * a); th = -5 + 1.8 * Math.sin(a * Math.PI * 4) * (1 - a); } }+ if (t >= Vg - 0.75) { ph = 10 * (1 - io(seg(t, Vg - 0.75, Vg - 0.05))); if (t >= Vg) th = -5 * (1 - land(seg(t, Vg, Vg + 0.6), 1.8)); }+ const bk2 = io(out), fixed = live && t >= Vg;+ tf(card, 160, 158, 1, lerp(th, -5, bk2).toFixed(2)); tf(brk, 12, 19, 1, lerp(ph, 10, bk2).toFixed(2));+ mend.setAttribute("opacity", (seg(t, Vg, Vg + 0.25) * (1 - out)).toFixed(2));+ tf(warn, 78, -24, (fixed ? 0 : (1 + 0.07 * Math.sin(now * 5)) * (live ? 1 : seg(t, 13.3, 13.6))).toFixed(3));+ tf(okm, 78, -24, (land(seg(t, Vg, Vg + 0.4), 2.4) * (1 - out)).toFixed(3));+ tf(sp, 284, 100, (land(seg(t, Vg + 0.3, Vg + 0.75), 2.6) * (1 - out) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ say(cap, !live || t < T0 ? "When an install fails, work up four steps" : t >= Vg ? "Stop at the first step that fixes it" : STEP[Math.min(goal, Math.floor((t - T0) / GAP))], 700);+ };+ },+ };++ /* ==== ADD NEW SCENES ABOVE THIS LINE ==== */++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/office.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/office.pngnew file mode 100644index 00000000..5d1a2239Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/office.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.pngnew file mode 100644index 00000000..edd4c53eBinary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/rooms.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/rooms.pngnew file mode 100644index 00000000..95a65f88Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/rooms.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/signin.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/signin.pngnew file mode 100644index 00000000..75cb6263Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/signin.png differdiff --git a/docs/blog/assets/media/how-to-install-and-use-munder-difflin/terminal.png b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/terminal.pngnew file mode 100644index 00000000..0a5c6e42Binary files /dev/null and b/docs/blog/assets/media/how-to-install-and-use-munder-difflin/terminal.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/ask-me-056.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/ask-me-056.pngnew file mode 100644index 00000000..2da60968Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/ask-me-056.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/askme.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/askme.pngnew file mode 100644index 00000000..5c9e159cBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/askme.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/brief.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/brief.pngnew file mode 100644index 00000000..b377d970Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/brief.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/hero.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/hero.pngindex 6eb3ee69..f2166cc6 100644Binary files a/docs/blog/assets/media/your-first-hour-with-munder-difflin/hero.png and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/hero.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/hour.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/hour.pngnew file mode 100644index 00000000..40dd9497Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/hour.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.css b/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.cssnew file mode 100644index 00000000..859374ca--- /dev/null+++ b/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.css@@ -0,0 +1,13 @@+/* Bright coded scenes for "How to Install Munder Difflin 0.5.6". Pairs with motion.js. */+.prose figure.mg { margin: 42px 0; }+.prose figure.mg img, .mg .mg-svg { display: block; width: 100%; height: auto; border: 1px solid #1A1320; border-radius: 18px; background: #fff; }+.mg figcaption { font: 500 13px/1.5 "Space Grotesk", system-ui, sans-serif; color: #6B5878; padding-top: 10px; text-align: center; }+.mg .mg-svg { touch-action: manipulation; -webkit-tap-highlight-color: transparent; }+.mg .mg-svg text { -webkit-user-select: none; user-select: none; pointer-events: none; }+.mg .mg-hit { cursor: pointer; outline: none; }+.mg [role="slider"] { cursor: grab; touch-action: none; }+.mg .mg-ring { stroke-opacity: 0; transition: stroke-opacity 0.15s; }+.mg .mg-hit:focus-visible .mg-ring { stroke-opacity: 1; }+.mg [role="slider"]:focus-visible circle:nth-child(2) { stroke-width: 5; }+@media (hover: hover) { .mg .mg-hit:hover .mg-ring { stroke-opacity: 0.5; } }+@media (prefers-reduced-motion: reduce) { .mg .mg-ring { transition: none; } }diff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.js b/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.jsnew file mode 100644index 00000000..aa644935--- /dev/null+++ b/docs/blog/assets/media/your-first-hour-with-munder-difflin/motion.js@@ -0,0 +1,1247 @@+/* Bright coded scenes for "Your First Hour With Munder Difflin 0.5.6". Vanilla JS, no library.+ <figure class="mg" data-scene="NAME"> holds a still image; this script swaps in a live SVG.+ Plays only in view, honours reduced motion, "?mgstill" freezes each scene on its still frame.++ TO ADD A SCENE: write SC.NAME = { dur, still, build(svg, api) { ...; return (t, now) => {} } } just above+ the line marked "ADD NEW SCENES ABOVE THIS LINE". Change nothing else in this file.+ - The drawing is 800 by 600 units (shown 1600 by 1200). dur is the loop in seconds, still is the frame for NAME.png.+ - build draws everything once and returns the frame function: t is the loop time, now never wraps (idle motion).+ - api.seek(s) jumps the loop to second s (a replay). api.now() is the clock. api.poke() redraws after a tap.+ - Helpers: el, txt, head (heading plus tap hint), strip and say (the caption line), button (tap and keyboard),+ pill, star, fit, mv (marks a group that moves, for the measurement tests), seg, io, oc, bk, land, lerp, tf. */+(() => {+ const NS = "http://www.w3.org/2000/svg";+ const INK = "#1A1320", YEL = "#FFCA54", BLUE = "#6C8EF5", LIL = "#B69CFF", MINT = "#7FD8AE", SKY = "#9AD8FF", SOFT = "#F1ECF9", MUTE = "#7A6A88";+ const SANS = '"Space Grotesk", system-ui, sans-serif', MONO = '"JetBrains Mono", ui-monospace, monospace';+ const el = (t, a = {}, p) => { const n = document.createElementNS(NS, t); for (const k in a) n.setAttribute(k, a[k]); if (p) p.appendChild(n); return n; };+ const txt = (p, x, y, s, a = {}) => { const n = el("text", Object.assign({ x, y, "font-family": SANS, "font-size": 24, "font-weight": 600, fill: INK }, a), p); n.textContent = s; return n; };+ const cl = (v, a = 0, b = 1) => Math.max(a, Math.min(b, v));+ const seg = (t, a, b) => cl((t - a) / (b - a));+ const oc = (x) => 1 - Math.pow(1 - x, 3);+ const io = (x) => (x < 0.5 ? 4 * x * x * x : 1 - Math.pow(-2 * x + 2, 3) / 2);+ const tf = (n, x, y, s = 1, r = 0) => n.setAttribute("transform", `translate(${x} ${y}) rotate(${r}) scale(${s})`);+ const button = (g, label, fn) => {+ g.setAttribute("role", "button"); g.setAttribute("tabindex", "0"); g.setAttribute("aria-label", label); g.classList.add("mg-hit");+ g.addEventListener("click", fn);+ g.addEventListener("keydown", (e) => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); fn(); } });+ };+ const strip = (svg, y, w = 680) => {+ el("rect", { x: (800 - w) / 2, y, width: w, height: 48, rx: 24, fill: SOFT }, svg);+ return txt(svg, 400, y + 31, "", { "text-anchor": "middle", "font-size": 20, "font-weight": 500 });+ };+ const lerp = (a, b, k) => a + (b - a) * k;+ /* ease that lands a little past the mark and settles; c sets how far past */+ const bk = (x, c) => 1 + (c + 1) * Math.pow(x - 1, 3) + c * Math.pow(x - 1, 2);+ const land = (k, c = 2.2) => cl(bk(k, c), 0, 1.5);+ /* scene heading on the left, the tap hint on the right; the heading shrinks if the two would touch */+ const head = (svg, s, hint) => {+ const a = txt(svg, 44, 62, s, { "font-size": 32, "font-weight": 700 });+ const b = txt(svg, 756, 60, hint, { "text-anchor": "end", "font-size": 20, fill: MUTE, "font-weight": 500 });+ const room = 756 - b.getComputedTextLength() - 28 - 44, w = a.getComputedTextLength();+ if (w > room) a.setAttribute("font-size", Math.floor((32 * room) / w));+ };+ const star = (p, x, y, r, fill) => el("path", { d: `M0 ${-r} Q0 0 ${r} 0 Q0 0 0 ${r} Q0 0 ${-r} 0 Q0 0 0 ${-r} Z`, fill, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", transform: `translate(${x} ${y})` }, p);+ /* shrink a label until it fits the room it has */+ const fit = (n, room) => { const w = n.getComputedTextLength(), f = parseFloat(n.getAttribute("font-size")); if (w > room) n.setAttribute("font-size", Math.floor((f * room * 10) / w) / 10); return n; };+ /* set the caption strip, at 20 unless the line is too long for the strip */+ const say = (n, s, room = 690) => { if (n._s === s) return; n._s = s; n.textContent = s; n.setAttribute("font-size", 20); fit(n, room); };+ /* a status pill sized to its word, with or without a tick */+ const pill = (p, s, fill, tick) => {+ const g = el("g", {}, p), r = el("rect", { y: -18, height: 36, rx: 18, fill, stroke: INK, "stroke-width": 2.5 }, g);+ const w = txt(g, tick ? 12 : 0, 7, s, { "text-anchor": "middle", "font-size": 20, "font-weight": 700 }).getComputedTextLength() + (tick ? 54 : 34);+ r.setAttribute("x", -w / 2); r.setAttribute("width", w);+ if (tick) el("path", { d: `M${-w / 2 + 15} 1 l6 6 l10 -13`, fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ return g;+ };+ /* mark a group as moving, so the measurement tests can see when it crosses a label */+ const mv = (g) => { g.setAttribute("data-mv", ""); return g; };++ const SC = {};++ /* ---- title card, 16 by 9: a stopwatch whose hand goes once round, one hour, and then it rings ---- */+ SC.hero = {+ w: 960, h: 540, dur: 4, still: 2.6,+ build(svg) {+ txt(svg, 66, 92, "From first sign in to a working office", { "font-size": 25, "font-weight": 500, fill: MUTE });+ const L = ["Your First Hour", "With Munder Difflin", "0.5.6"].map((s, i) => txt(svg, 62, 176 + i * 74, s, { "font-size": 76, "font-weight": 700 }));+ const w = Math.max(...L.map((n) => n.getComputedTextLength()));+ if (w > 556) L.forEach((n) => n.setAttribute("font-size", Math.floor((76 * 556) / w)));+ el("path", { d: `M68 344 H${Math.round(56 + L[2].getComputedTextLength())}`, stroke: YEL, "stroke-width": 14, "stroke-linecap": "round" }, svg);+ const pt = txt(svg, 0, 419, "Onboarding guide", { "text-anchor": "middle", "font-size": 25, "font-weight": 700 }), pw = Math.round(pt.getComputedTextLength() + 64);+ svg.insertBefore(el("rect", { x: 62, y: 382, width: pw, height: 56, rx: 28, fill: SOFT, stroke: INK, "stroke-width": 3 }), pt);+ pt.setAttribute("x", 62 + pw / 2);+ el("path", { d: "M70 494 C190 468 300 518 420 490 S560 492 620 478", stroke: LIL, "stroke-width": 5, fill: "none", "stroke-linecap": "round", "stroke-dasharray": "1 16" }, svg);+ el("circle", { cx: 70, cy: 494, r: 11, fill: MINT, stroke: INK, "stroke-width": 3 }, svg);+ /* the stopwatch: a ring on top, two side buttons, the crown, a face with sixty marks */+ const c = el("g", {}, svg);+ el("circle", { cx: 0, cy: -172, r: 15, fill: "none", stroke: INK, "stroke-width": 4 }, c);+ [-1, 1].forEach((s) => el("rect", { x: -12, y: -142, width: 24, height: 26, rx: 6, fill: LIL, stroke: INK, "stroke-width": 3, transform: `rotate(${s * 40})` }, c));+ el("rect", { x: -11, y: -136, width: 22, height: 22, fill: "#fff", stroke: INK, "stroke-width": 3 }, c);+ const crown = el("rect", { x: -24, y: -160, width: 48, height: 26, rx: 7, fill: YEL, stroke: INK, "stroke-width": 3 }, c);+ el("circle", { r: 118, fill: BLUE, stroke: INK, "stroke-width": 3 }, c);+ el("circle", { r: 98, fill: "#fff", stroke: INK, "stroke-width": 3 }, c);+ const LEN = 2 * Math.PI * 62;+ const ring = el("circle", { r: 62, fill: "none", stroke: YEL, "stroke-width": 22, transform: "rotate(-90)" }, c);+ el("circle", { r: 88, fill: "none", stroke: MUTE, "stroke-width": 7, "stroke-dasharray": `2 ${((2 * Math.PI * 88) / 60 - 2).toFixed(3)}`, "stroke-dashoffset": 1 }, c);+ for (let i = 0; i < 12; i++) el("path", { d: "M0 -76 V-92", stroke: INK, "stroke-width": 4.5, "stroke-linecap": "round", transform: `rotate(${i * 30})` }, c);+ const hand = el("path", { d: "M0 16 V-78", stroke: INK, "stroke-width": 5, "stroke-linecap": "round" }, c);+ el("circle", { r: 10, fill: INK }, c);+ el("circle", { r: 4, fill: YEL }, c);+ const ding = el("path", { d: [-1, 1].map((s) => `M${s * 128} -42 q${s * 12} 42 0 84 M${s * 142} -60 q${s * 14} 60 0 120`).join(" "), fill: "none", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, c);+ star(svg, 646, 132, 15, YEL); star(svg, 912, 470, 17, MINT); star(svg, 662, 456, 12, SKY);+ return (t) => {+ const k = io(seg(t, 0.35, 2.3)), out = io(seg(t, 3.5, 3.95)), tau = t - 2.3, ring2 = tau > 0 && tau < 1.3 ? Math.exp(-2.5 * tau) : 0;+ ring.setAttribute("stroke-dasharray", `${(LEN * k * (1 - out)).toFixed(2)} ${(LEN + 4).toFixed(2)}`);+ hand.setAttribute("transform", `rotate(${(360 * k * (1 - out) + (tau > 0 ? 6 * Math.exp(-5 * tau) * Math.sin(20 * tau) : 0)).toFixed(2)})`);+ crown.setAttribute("transform", `translate(0 ${(5 * Math.sin(Math.PI * seg(t, 0.1, 0.35))).toFixed(2)})`);+ c.setAttribute("transform", `translate(${(794 + 2.2 * ring2 * Math.sin(tau * 60)).toFixed(2)} 296)`);+ ding.setAttribute("opacity", (seg(t, 2.3, 2.4) * (1 - seg(t, 3.3, 3.55))).toFixed(2));+ ding.setAttribute("transform", `scale(${(1 + 0.04 * Math.sin(Math.max(0, tau) * 26)).toFixed(3)})`);+ };+ },+ };++ /* ---- hour: a bus route with six stops, one for each part of the hour; the bus takes the hairpin a little fast ---- */+ SC.hour = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Your first hour", "tap a stop");+ const X0 = 122, XE = 634, R1 = 222, R2 = 406, RAD = 92, RUN = XE - X0, ARC = Math.PI * RAD, LEN = 2 * RUN + ARC;+ const S = [0, 216, 432, RUN + ARC + 80, RUN + ARC + 296, LEN];+ const LIT = [0.8, 2.6, 4.6, 7.3, 9.3, 11.3], DEP = [1.6, 3.5, 5.5, 8.2, 10.2];+ const ST = [["Minute 0", ["sign in"]], ["Minute 10", ["a first job"]], ["Minute 20", ["a small team"]], ["Minute 35", ["work that starts", "without you"]], ["Minute 45", ["temps and tokens"]], ["Minute 50", ["the Pro screens"]]];+ const SAY = ["Minute 0: sign in", "Minute 10: give your orchestrator one real job", "Minute 20: hire a small team", "Minute 35: set up work that starts without you", "Minute 45: temps and tokens", "Minute 50: learn the Pro screens"];+ /* where the bus is at distance s along the route, and how far round the hairpin it has come */+ const at = (s) => {+ if (s <= RUN) return [X0 + s, R1, 0];+ if (s <= RUN + ARC) { const a = (s - RUN) / RAD; return [XE + RAD * Math.sin(a), R1 + RAD - RAD * Math.cos(a), a]; }+ return [XE - (s - RUN - ARC), R2, Math.PI];+ };+ /* trees and a pond, because it is a map */+ const trees = [[700, 164, 17], [744, 160, 23], [668, 494, 15], [710, 498, 23], [752, 492, 17]].map(([x, y, r]) => {+ const g = el("g", {}, svg);+ el("path", { d: `M${x} ${y} V${y - r - 10}`, stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round" }, g);+ el("circle", { cx: x, cy: y - r - 14, r, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ return { g, x, y };+ });+ el("ellipse", { cx: 662, cy: 314, rx: 36, ry: 24, fill: SKY, stroke: INK, "stroke-width": 2.5 }, svg);+ const rip = el("ellipse", { cx: 662, cy: 314, fill: "none", stroke: "#fff", "stroke-width": 3 }, svg);+ /* the road, and the part of it the bus has already driven */+ const D = `M${X0} ${R1} H${XE} A${RAD} ${RAD} 0 0 1 ${XE} ${R2} H${X0}`;+ el("path", { d: D, fill: "none", stroke: INK, "stroke-width": 18, "stroke-linecap": "round" }, svg);+ el("path", { d: D, fill: "none", stroke: "#fff", "stroke-width": 12, "stroke-linecap": "round" }, svg);+ const done = el("path", { d: D, fill: "none", stroke: BLUE, "stroke-width": 12, "stroke-linecap": "round" }, svg);+ /* skid marks on the outside of the hairpin */+ const A0 = 0.95, A1 = 2.2;+ const skid = [RAD + 16, RAD + 23].map((r) => {+ const p = (a) => `${(XE + r * Math.sin(a)).toFixed(1)} ${(R1 + RAD - r * Math.cos(a)).toFixed(1)}`;+ return { n: el("path", { d: `M${p(A0)} A${r} ${r} 0 0 1 ${p(A1)}`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg), L: r * (A1 - A0) };+ });+ /* the six stops; each one is a button that sends the bus there */+ let ov = null, cur = 0;+ const send = (i) => { ov = { s: cur, to: S[i], t0: api.now() }; api.seek(LIT[i] - 0.8); };+ const stops = ST.map(([m, nm], i) => {+ const [x, y] = at(S[i]), g = el("g", {}, svg);+ el("rect", { x: x - 102, y: y - 112, width: 204, height: nm.length > 1 ? 192 : 162, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const p = pill(g, m, SOFT);+ const dot = el("circle", { cx: x, cy: y, r: 12, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ const names = nm.map((s, j) => txt(g, x, y + 44 + j * 27, s, { "text-anchor": "middle", "font-size": 22, fill: MUTE }));+ button(g, `Send the bus to ${m.toLowerCase()}: ${nm.join(" ")}`, () => send(i));+ return { x, y, p, r: p.firstChild, dot, names };+ });+ /* the bus, side on; it stays upright and turns round in the hairpin */+ const bus = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -35, y: -46, width: 70, height: 34, rx: 9, fill: YEL, stroke: INK, "stroke-width": 3 }, bus);+ [-28, -10].forEach((x) => el("rect", { x, y: -40, width: 14, height: 12, rx: 3, fill: "#fff", stroke: INK, "stroke-width": 2 }, bus));+ el("rect", { x: 8, y: -40, width: 21, height: 12, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2 }, bus);+ el("path", { d: "M-35 -21 H35", stroke: INK, "stroke-width": 2 }, bus);+ const wh = [-18, 18].map((x) => {+ const w = el("g", {}, bus);+ el("circle", { r: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, w);+ el("path", { d: "M-4 0 H4", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, w);+ return { w, x };+ });+ const sp = el("g", { "pointer-events": "none" }, svg); star(sp, 0, 0, 13, YEL);+ const cap = strip(svg, 508, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.45), vis = 1 - out, home = t >= 13.45;+ let s = home ? 0 : LEN, last = 0, turn = 1;+ for (let i = 0; i < 6; i++) if (t >= LIT[i]) last = i;+ if (!home) for (let i = 0; i < 5; i++) if (t < LIT[i + 1]) { s = lerp(S[i], S[i + 1], io(seg(t, DEP[i], LIT[i + 1]))); break; }+ /* after a tap the bus drives from where it was to the stop that was tapped, then turns to face the route again */+ if (ov) {+ const k = (now - ov.t0) / 0.8;+ if (k < 0 || k >= 1) ov = null;+ else { const d = Math.sign(ov.to - ov.s) || 1; s = lerp(ov.s, s, io(k)); turn = lerp(d, 1, io(seg(k, 0.72, 1))); }+ }+ cur = s;+ const [x, y, a] = at(s);+ const face = lerp(1, -1, io(seg(a, 1.15, 2.0))) * turn, lean = 13 * Math.sin(a);+ const ta = t - LIT[last], hop = !ov && ta > 0 && ta < 1.2 ? 4 * Math.exp(-6 * ta) * Math.abs(Math.sin(15 * ta)) : 0;+ const idle = t > 11.9 && !home ? 0.8 * Math.sin(now * 14) : 0;+ bus.setAttribute("transform", `translate(${x.toFixed(2)} ${(y - 8 - hop + idle).toFixed(2)}) rotate(${lean.toFixed(2)}) scale(${face.toFixed(3)} 1)`);+ bus.setAttribute("opacity", (home ? seg(t, 13.5, 13.75) : vis).toFixed(2));+ wh.forEach((w) => w.w.setAttribute("transform", `translate(${w.x} -8) rotate(${((s / 8) * 57.3).toFixed(1)})`));+ done.setAttribute("stroke-dasharray", `${s.toFixed(1)} ${LEN + 40}`);+ done.setAttribute("opacity", vis.toFixed(2));+ const sk = seg(s, RUN + RAD * (A0 + 0.2), RUN + RAD * (A1 + 0.35));+ skid.forEach((m) => { m.n.setAttribute("stroke-dasharray", `${(m.L * sk).toFixed(1)} ${(m.L + 20).toFixed(1)}`); m.n.setAttribute("opacity", sk > 0 ? vis.toFixed(2) : 0); });+ stops.forEach((st, i) => {+ const lit = t >= LIT[i] && !home && out < 0.5, bump = Math.sin(Math.PI * seg(t, LIT[i], LIT[i] + 0.4)) * vis;+ tf(st.p, st.x, st.y - 84, (1 + 0.12 * bump).toFixed(3));+ st.r.setAttribute("fill", lit ? YEL : SOFT);+ st.dot.setAttribute("fill", lit ? YEL : "#fff");+ st.dot.setAttribute("r", (12 + 3 * bump).toFixed(2));+ st.names.forEach((n) => n.setAttribute("fill", lit ? INK : MUTE));+ });+ tf(sp, 182, 366, (land(seg(t, 11.5, 11.9), 2.6) * (home ? 0 : vis) * (1 + 0.12 * Math.sin(now * 3))).toFixed(3), (now * 20).toFixed(1));+ trees.forEach((tr, i) => tr.g.setAttribute("transform", `rotate(${(1.6 * Math.sin(now * 1.4 + i * 1.3)).toFixed(2)} ${tr.x} ${tr.y})`));+ const rp = (now * 0.4 + 0.35) % 1;+ rip.setAttribute("rx", (6 + 24 * rp).toFixed(1)); rip.setAttribute("ry", (4 + 15 * rp).toFixed(1)); rip.setAttribute("opacity", (1 - rp).toFixed(2));+ say(cap, t >= LIT[0] && !home && out < 0.5 ? SAY[last] : "The first hour as a route", 700);+ };+ },+ };++ /* ---- brief: a long toaster with two slots; a vague note comes back as a question, a full brief as three tasks ---- */+ SC.brief = {+ dur: 14, still: 12,+ build(svg, api) {+ head(svg, "Give your orchestrator a first job", "tap a note");+ const XA = 212, XB = 588, TOP = 388;+ const clip = el("clipPath", { id: "mg-brief-slots" }, el("defs", {}, svg));+ el("rect", { x: 0, y: 0, width: 800, height: TOP + 1 }, clip);+ /* the two notepads; each is a button that drops its note again */+ const pad = (x, tag, tagFill, lines, label, at) => {+ const g = el("g", {}, svg);+ el("rect", { x: x - 152, y: 80, width: 304, height: 190, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ el("rect", { x: x - 126, y: 134, width: 264, height: 126, rx: 8, fill: SOFT, stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: x - 132, y: 128, width: 264, height: 126, rx: 8, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: x - 132, y: 128, width: 264, height: 20, rx: 8, fill: LIL, stroke: INK, "stroke-width": 3 }, g);+ [0, 1, 2].forEach((j) => el("path", { d: `M${x - 114} ${189 + j * 28} H${x + 114}`, stroke: LIL, "stroke-width": 1.5 }, g));+ lines.forEach((s, j) => {+ el("circle", { cx: x - 108, cy: 175 + j * 28, r: 4.5, fill: BLUE }, g);+ txt(g, x - 96, 182 + j * 28, s, { "font-size": 21, "font-weight": 600 });+ });+ tf(pill(g, tag, tagFill), x, 104);+ button(g, label, () => api.seek(at));+ };+ pad(XA, "Weak", SOFT, ["clean up my project"], "Drop the weak note again", 0.7);+ pad(XB, "Better", MINT, ["the outcome", "the folder", "what done looks like"], "Drop the better note again", 5.0);+ /* the back of the toaster: the two slots, which glow while it works */+ const back = el("g", {}, svg);+ const slots = [XA, XB].map((x) => el("rect", { x: x - 80, y: TOP - 9, width: 160, height: 18, rx: 9, fill: INK, stroke: INK, "stroke-width": 3 }, back));+ /* what goes in and what comes out, clipped at the top of the toaster */+ const lay = el("g", { "clip-path": "url(#mg-brief-slots)" }, svg);+ const slip = (n) => {+ const g = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -36, y: -26, width: 72, height: 52, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 3 }, g);+ el("rect", { x: -36, y: -26, width: 72, height: 11, rx: 5, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: n === 1 ? "M-22 4 H16" : "M-22 -2 H20 M-22 8 H12 M-22 18 H22", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ };+ const SA = slip(1), SB = slip(3);+ const Q = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -52, y: -116, width: 104, height: 140, rx: 10, fill: SOFT, stroke: INK, "stroke-width": 3 }, Q);+ txt(Q, 0, -38, "?", { "text-anchor": "middle", "font-size": 70, "font-weight": 700, fill: MUTE });+ const tasks = [SKY, LIL, YEL].map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, lay));+ el("rect", { x: -38, y: -98, width: 76, height: 94, rx: 8, fill: c, stroke: INK, "stroke-width": 3 }, g);+ const box = el("rect", { x: -26, y: -86, width: 24, height: 24, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ const tick = el("path", { d: "M-20 -74 l4.5 5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ el("path", { d: "M6 -74 H26 M-26 -48 H26 M-26 -33 H26 M-26 -18 H10", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, box, tick };+ });+ /* the front of the toaster */+ const front = el("g", {}, svg);+ const BOT = TOP + 102;+ el("path", { d: `M132 ${TOP} H668 A40 40 0 0 1 708 ${TOP + 40} V${BOT - 14} A14 14 0 0 1 694 ${BOT} H106 A14 14 0 0 1 92 ${BOT - 14} V${TOP + 40} A40 40 0 0 1 132 ${TOP} Z`, fill: LIL, stroke: INK, "stroke-width": 3 }, front);+ el("path", { d: `M136 ${TOP + 20} H664`, stroke: "#fff", "stroke-width": 4, "stroke-linecap": "round" }, front);+ [122, 644].forEach((x) => el("rect", { x, y: BOT, width: 34, height: 8, rx: 3, fill: INK }, front));+ txt(front, 400, TOP + 66, "your orchestrator", { "text-anchor": "middle", "font-size": 27, "font-weight": 700 });+ el("circle", { cx: 150, cy: TOP + 58, r: 20, fill: "#fff", stroke: INK, "stroke-width": 3 }, front);+ const dial = el("path", { d: "M0 0 V-13", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round" }, front);+ el("rect", { x: 648, y: TOP + 32, width: 12, height: 58, rx: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, front);+ const lever = el("rect", { x: 634, width: 40, height: 16, rx: 8, fill: YEL, stroke: INK, "stroke-width": 3 }, front);+ /* the vague note is still smoking a little */+ const puff = [0, 1].map(() => el("path", { d: "M0 0 q-9 -11 0 -22 t0 -22", fill: "none", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, svg));+ const cap = strip(svg, 508, 740);+ const PA = 3.3, PB = 7.6;+ return (t) => {+ const out = seg(t, 13.2, 13.45), vis = 1 - out, sink = 30 * io(seg(t, 13.2, 13.5));+ const cookA = t >= 2.05 && t < PA, cookB = t >= 6.35 && t < PB, cook = cookA || cookB;+ /* the slips: each pops out under its pad and drops into its slot */+ [[SA, XA, 0.9], [SB, XB, 5.2]].forEach(([g, x, a]) => {+ const f = io(seg(t, a + 0.45, a + 1.1));+ tf(g, x, (292 + (TOP + 70 - 292) * f).toFixed(2), land(seg(t, a, a + 0.3), 2).toFixed(3), (-8 * Math.sin(Math.PI * f)).toFixed(2));+ g.setAttribute("opacity", t >= a && t < a + 1.15 ? 1 : 0);+ });+ /* the toaster works, then pops */+ const lv = Math.max(seg(t, 1.95, 2.1) * (1 - seg(t, PA, PA + 0.06)), seg(t, 6.25, 6.4) * (1 - seg(t, PB, PB + 0.06)));+ const ha = t - PA, hb = t - PB, hp = (v) => (v > 0 && v < 1 ? 5 * Math.exp(-7 * v) * Math.abs(Math.sin(16 * v)) : 0);+ const jig = cook ? 1.3 * Math.sin(t * 60) : 0, lift = hp(ha) + hp(hb);+ [back, front].forEach((g) => g.setAttribute("transform", `translate(${jig.toFixed(2)} ${(-lift).toFixed(2)})`));+ lever.setAttribute("y", (TOP + 30 + 46 * lv).toFixed(2));+ slots[0].setAttribute("fill", cookA ? YEL : INK);+ slots[1].setAttribute("fill", cookB ? YEL : INK);+ const dk = cookA ? seg(t, 2.1, PA) : cookB ? seg(t, 6.4, PB) : 1;+ dial.setAttribute("transform", `translate(150 ${TOP + 58}) rotate(${(150 * (1 - dk)).toFixed(1)})`);+ /* the vague note comes back as one grey card with a question mark */+ const qa = land(seg(t, PA, PA + 0.42), 1.7), qw = t - PA - 0.4;+ tf(Q, XA, (TOP + 120 * (1 - qa) + sink).toFixed(2), 1, (qw > 0 ? 7 * Math.exp(-4 * qw) * Math.sin(13 * qw) : 0).toFixed(2));+ Q.setAttribute("opacity", t >= PA ? vis.toFixed(2) : 0);+ puff.forEach((c, j) => {+ const p0 = (t - PA - 0.1) * 0.4 - j * 0.5, p = p0 % 1;+ c.setAttribute("transform", `translate(${(XA + 64 + 8 * j + 4 * Math.sin(p * 5)).toFixed(1)} ${(TOP - 10 - 58 * p).toFixed(1)}) scale(${(0.7 + 0.45 * p).toFixed(3)})`);+ c.setAttribute("opacity", p0 < 0 ? 0 : (cl(p * 5) * (1 - p) * vis).toFixed(2));+ });+ /* the full brief comes back as three task cards that hop apart, and each gets its tick */+ const ta = land(seg(t, PB, PB + 0.4), 1.6), spr = io(seg(t, PB + 0.45, PB + 1.05));+ tasks.forEach((k, i) => {+ const hop = i === 1 ? 0 : 16 * Math.sin(Math.PI * spr);+ tf(k.g, (XB + (i - 1) * 86 * spr).toFixed(2), (TOP + 104 * (1 - ta) - hop + sink).toFixed(2), 1, ((i - 1) * 10 * Math.sin(Math.PI * spr)).toFixed(2));+ k.g.setAttribute("opacity", t >= PB ? vis.toFixed(2) : 0);+ const tk = seg(t, 9.1 + 0.45 * i, 9.4 + 0.45 * i);+ k.tick.setAttribute("stroke-dasharray", `${(21 * tk).toFixed(2)} 30`);+ k.tick.setAttribute("opacity", tk > 0 ? 1 : 0);+ k.box.setAttribute("fill", tk > 0.5 ? MINT : "#fff");+ });+ say(cap, t >= 13.2 || t < PA ? "A vague brief gets a vague job" : t < 5.2 ? "A vague one turns into a question" : t < PB ? "The outcome, the folder, and what done looks like" : "Turns into clear tasks", 700);+ };+ },+ };++ /* ---- team: three office chairs roll in, one per hire; a change goes from the builder to the reviewer and back with a tick ---- */+ SC.team = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Hire a small team", "tap the builder");+ const XS = [150, 400, 650], FL = 400, COL = [SKY, LIL, MINT], PR = [0.7, 2.6, 4.3], IN = [0.85, 2.75, 4.45];+ const ROLE = ["builder", "reviewer", "helper"], ENG = ["on Claude Code", "on Codex", "on a free engine"];+ el("path", { d: `M44 ${FL + 2} H756`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ /* three open seats, drawn as dotted outlines */+ const ghost = XS.map((x) => {+ const g = el("g", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round" }, svg);+ el("rect", { x: x - 50, y: FL - 214, width: 100, height: 92, rx: 22 }, g);+ el("rect", { x: x - 56, y: FL - 104, width: 112, height: 22, rx: 11 }, g);+ el("path", { d: `M${x} ${FL - 80} V${FL - 14} M${x - 38} ${FL - 6} H${x + 38}` }, g);+ return g;+ });+ /* under each seat: its role, and the engine it runs on once hired */+ const zone = XS.map((x, i) => {+ const g = el("g", {}, svg);+ if (i === 0) el("rect", { x: x - 112, y: 170, width: 224, height: 332, rx: 16, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ const slot = el("rect", { x: x - 96, y: 458, width: 192, height: 36, rx: 18, fill: "none", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, g);+ const role = txt(g, x, 438, ROLE[i], { "text-anchor": "middle", "font-size": 26, "font-weight": 700, fill: MUTE });+ const p = pill(g, ENG[i], COL[i]);+ return { slot, role, p };+ });+ button(zone[0].role.parentNode, "Send another change from the builder to the reviewer", () => api.seek(6.5));+ /* the chairs */+ const chairs = COL.map((c) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg)), top = el("g", {}, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, top);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, top);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, top);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, top);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, top);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, top);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, top);+ return { g, top };+ });+ /* the helper's papers, in a heap until they are tidied */+ const pile = [[-14, -16], [2, 9], [12, 22]].map(([dx, r]) => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-9 -10 H9 M-9 0 H9 M-9 10 H3", stroke: MUTE, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return { g, dx, r };+ });+ /* the change: a sheet that goes to the reviewer and comes back with a tick */+ const sheet = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -22, y: -28, width: 44, height: 56, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, sheet);+ const sl = [-12, 0, 12].map((y, j) => el("path", { d: `M-12 ${y} H${j === 2 ? 4 : 12}`, stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, sheet));+ const ok = el("g", {}, sheet);+ el("circle", { r: 12, fill: MINT, stroke: INK, "stroke-width": 2.5 }, ok);+ el("path", { d: "M-5.5 0 l4 4.5 l7.5 -9", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, ok);+ /* the button you press, and the pointer pressing it */+ const B = el("g", {}, svg);+ el("rect", { x: -100, y: -24, width: 200, height: 48, rx: 14, fill: YEL, stroke: INK, "stroke-width": 3 }, B);+ txt(B, 0, 8, "Add an agent", { "text-anchor": "middle", "font-size": 23, "font-weight": 700 });+ const rip = el("circle", { fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, svg);+ const cap = strip(svg, 508, 740);+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, idle = t > 11.8 && t < 13.2 ? 1 : 0;+ const press = Math.max(...PR.map((a) => bump(t, a)));+ tf(B, 400, 122, (1 - 0.06 * press + 0.012 * idle * Math.sin(now * 3)).toFixed(3));+ const lastP = PR.filter((a) => t >= a).pop(), rk = lastP === undefined ? 1 : seg(t, lastP, lastP + 0.6);+ rip.setAttribute("cx", 474); rip.setAttribute("cy", 138);+ rip.setAttribute("r", (8 + 30 * oc(rk)).toFixed(1)); rip.setAttribute("opacity", (rk < 1 ? 1 - rk : 0).toFixed(2));+ ptr.setAttribute("transform", `translate(${(474 + 2 * press).toFixed(2)} ${(138 + 3 * press + 2 * idle * Math.sin(now * 2)).toFixed(2)})`);+ /* each press rolls a chair in from the right; the third one arrives still spinning */+ chairs.forEach((ch, i) => {+ const k = seg(t, IN[i], IN[i] + 1.05), x = lerp(880, XS[i], land(k, 1.15)), st = t - IN[i] - 1.05;+ const lean = 3 * Math.sin(Math.PI * k) - (st > 0 ? 3 * Math.exp(-5 * st) * Math.sin(14 * st) : 0);+ tf(ch.g, x.toFixed(2), FL, 1, lean.toFixed(2));+ ch.g.setAttribute("opacity", t < IN[i] ? 0 : vis.toFixed(2));+ const sx = i === 2 ? Math.cos(4 * Math.PI * (1 - oc(seg(t, IN[2], IN[2] + 1.9)))) : 1;+ ch.top.setAttribute("transform", `scale(${sx.toFixed(3)} 1)`);+ ghost[i].setAttribute("opacity", Math.max(1 - seg(t, IN[i] + 0.6, IN[i] + 1.0), out).toFixed(2));+ const z = zone[i], on = t >= IN[i] + 1.0 && t < 13.2, pk = land(seg(t, IN[i] + 1.05, IN[i] + 1.4), 2.4) * vis;+ z.role.setAttribute("fill", on ? INK : MUTE);+ tf(z.p, XS[i], 476, pk.toFixed(3));+ z.slot.setAttribute("opacity", pk > 0.6 ? 0 : 1);+ });+ /* the change goes over, is read line by line, gets its tick and comes back */+ const g1 = io(seg(t, 7.2, 8.2)), g2 = io(seg(t, 9.9, 10.9)), back = t >= 9.9, k = back ? g2 : g1, lw = t - 10.9;+ const sx = lerp(XS[0], XS[1], back ? 1 - g2 : g1), sy = FL - 132 - 70 * Math.sin(Math.PI * k);+ const rot = (back ? -12 : 12) * Math.sin(Math.PI * k) + (lw > 0 ? 6 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0);+ tf(sheet, sx.toFixed(2), sy.toFixed(2), (land(seg(t, 6.6, 6.95), 2.2) * vis).toFixed(3), rot.toFixed(2));+ sl.forEach((l, j) => l.setAttribute("stroke", t > 8.45 + 0.3 * j && t < 13.2 ? INK : MUTE));+ tf(ok, 20, -24, land(seg(t, 9.45, 9.8), 2.6).toFixed(3));+ /* the helper's heap lands, then squares itself into a neat stack */+ pile.forEach((p, j) => {+ const dk = land(seg(t, 6.0 + 0.12 * j, 6.35 + 0.12 * j), 1.6), td = io(seg(t, 11.0 + 0.1 * j, 11.6 + 0.1 * j));+ tf(p.g, (XS[2] + lerp(p.dx, (j - 1) * 3, td)).toFixed(2), (FL - 130 - 60 * (1 - dk) - 3 * j * td).toFixed(2), 1, lerp(p.r, 0, td).toFixed(2));+ p.g.setAttribute("opacity", t < 6.0 + 0.12 * j ? 0 : vis.toFixed(2));+ });+ say(cap, t < PR[0] || t >= 13.2 ? "Press Add an agent" : t < PR[1] ? "A builder on Claude Code, writing the feature" : t < PR[2] ? "A reviewer on Codex, reading every change the builder makes" : t < 6.6 ? "A helper on a free engine, for docs and tidying" : t < 11 ? "Reading every change the builder makes" : "A first team of three", 700);+ };+ },+ };++ /* ---- schedule: press New automation, choose Schedule, fill in four things and walk off; the sun comes up and the card fires without you ---- */+ SC.schedule = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "Work that starts without you", "tap the card");+ const X0 = 44, X1 = 500, Y0 = 88, Y1 = 474, HZ = 226, FL = 472, AX = 652, VX = 176, VW = 306, FY = [228, 288, 348, 420];+ const around = (n, cx, cy, s) => n.setAttribute("transform", `translate(${cx} ${cy}) scale(${s}) translate(${-cx} ${-cy})`);+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ const mini = (p, c) => {+ const g = el("g", {}, p);+ el("rect", { x: -9, y: -17, width: 18, height: 14, rx: 5, fill: c, stroke: INK, "stroke-width": 2 }, g);+ el("rect", { x: -11, y: -2, width: 22, height: 6, rx: 3, fill: c, stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M0 4 V11 M-8 13 H8", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ };+ /* the right side: the arc the sun climbs, the horizon, the floor and the agent's chair */+ el("path", { d: `M548 ${HZ} A104 104 0 0 1 756 ${HZ}`, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "2 10", "stroke-linecap": "round" }, svg);+ const cp = el("clipPath", { id: "mg-schedule-sky" }, el("defs", {}, svg));+ el("rect", { x: 500, y: 70, width: 300, height: HZ - 70 }, cp);+ const sun = mv(el("g", { "pointer-events": "none" }, el("g", { "clip-path": "url(#mg-schedule-sky)" }, svg)));+ const rays = el("g", {}, sun);+ for (let i = 0; i < 8; i++) el("path", { d: "M0 -35 V-46", stroke: INK, "stroke-width": 4, "stroke-linecap": "round", transform: `rotate(${i * 45})` }, rays);+ el("circle", { r: 25, fill: YEL, stroke: INK, "stroke-width": 3 }, sun);+ el("path", { d: `M524 ${HZ} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ el("path", { d: `M524 ${FL} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const ag = chair(svg, SKY, 0.78);+ /* the card: Automations, the New automation button, the two kinds, then the four fields */+ const card = el("g", {}, svg);+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 3 }, card);+ el("rect", { x: X0 - 9, y: Y0 - 9, width: X1 - X0 + 18, height: Y1 - Y0 + 18, rx: 24, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, card);+ button(card, "Fire the schedule again", () => api.seek(8.7));+ txt(card, 66, 128, "Automations", { "font-size": 24, "font-weight": 700 });+ const nb = el("g", {}, card);+ el("rect", { x: -98, y: -21, width: 196, height: 42, rx: 13, fill: YEL, stroke: INK, "stroke-width": 3 }, nb);+ txt(nb, 0, 7, "New automation", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ el("path", { d: `M${X0 + 2} 150 H${X1 - 2}`, stroke: SOFT, "stroke-width": 3 }, card);+ const k1 = pill(card, "Schedule", "#fff"), k2 = pill(card, "Webhook endpoint", "#fff");+ const w1 = parseFloat(k1.firstChild.getAttribute("width")), w2 = parseFloat(k2.firstChild.getAttribute("width")), KX = [66 + w1 / 2, 80 + w1 + w2 / 2];+ const names = ["Label", "When", "Sends to", "Message"].map((s, i) => txt(card, 66, FY[i] + 6, s, { "font-size": 18, "font-weight": 700, fill: MUTE }));+ const box = (y, h) => el("rect", { x: VX, y, width: VW, height: h, rx: 12, fill: "#fff", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, card);+ const bx = [box(FY[0] - 20, 40), box(FY[1] - 20, 40), box(FY[2] - 20, 40), box(FY[3] - 32, 66)];+ const wk = el("rect", { x: VX + VW / 2 + 4, y: FY[1] - 16, width: VW / 2 - 8, height: 32, rx: 9, fill: YEL, stroke: INK, "stroke-width": 2 }, card);+ el("path", { d: `M${VX + VW / 2} ${FY[1] - 13} V${FY[1] + 13}`, stroke: LIL, "stroke-width": 2.5 }, card);+ const segT = ["Interval", "Weekly"].map((s, i) => txt(card, VX + VW / 4 + (i * VW) / 2, FY[1] + 7, s, { "text-anchor": "middle", "font-size": 19, "font-weight": 600, fill: MUTE }));+ const vL = txt(card, VX + 16, FY[0] + 7, "", { "font-size": 19, "font-weight": 600 });+ const vM = [txt(card, VX + 16, FY[3] - 5, "", { "font-size": 19, "font-weight": 600 }), txt(card, VX + 16, FY[3] + 21, "", { "font-size": 19, "font-weight": 600 })];+ const caret = el("rect", { width: 2.5, height: 21, fill: INK }, card);+ const st = el("g", {}, card);+ mini(st, SKY).setAttribute("transform", "translate(14 2) scale(0.86)");+ txt(st, 34, 7, "an agent", { "font-size": 19, "font-weight": 700 });+ /* when it fires: a ripple off the card and a slip that flies to the chair */+ const rip = el("circle", { fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ const slip = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -19, y: -25, width: 38, height: 50, rx: 5, fill: "#fff", stroke: INK, "stroke-width": 3 }, slip);+ [-11, 0, 11].forEach((y, j) => el("path", { d: `M-10 ${y} H${j === 2 ? 3 : 10}`, stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, slip));+ const ok = el("g", {}, slip);+ el("circle", { r: 11, fill: MINT, stroke: INK, "stroke-width": 2.5 }, ok);+ el("path", { d: "M-5 0 l3.6 4 l7 -8.5", fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, ok);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 508, 740);+ const PR = [0.85, 2.35, 3.25, 5.25, 5.95, 6.45];+ const KP = [[0, 572, 296], [0.6, 478, 130], [1.6, 478, 130], [2.15, 66 + w1 - 12, 188], [2.7, 66 + w1 - 12, 188], [3.1, 470, 236], [4.6, 470, 236], [5.05, 470, 296], [5.5, 470, 296], [5.8, 470, 356], [6.15, 470, 356], [6.4, 470, 436], [7.9, 470, 436], [8.6, 870, 420], [10.8, 870, 318], [11.3, 742, 318], [11.9, 742, 318], [12.35, 870, 318], [13.4, 870, 318], [14, 572, 296]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ const typed = (n, s, t, a, b) => { const c = Math.round(s.length * seg(t, a, b)), v = t < 13.2 ? s.slice(0, c) : s; if (n._v !== v) { n._v = v; n.textContent = v; } return t > a && t < b + 0.15; };+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, live = t < 13.2;+ const press = Math.max(...PR.map((a) => bump(t, a)));+ tf(nb, 392, 120, (1 - 0.07 * bump(t, PR[0])).toFixed(3));+ [k1, k2].forEach((k, i) => { tf(k, KX[i], 182, land(seg(t, 1.05 + 0.12 * i, 1.4 + 0.12 * i), 2).toFixed(3)); k.setAttribute("opacity", vis.toFixed(2)); });+ k1.firstChild.setAttribute("fill", t >= PR[1] + 0.12 && live ? LIL : "#fff");+ /* the fields wake up once Schedule is chosen */+ bx.forEach((b, i) => { const on = t >= 2.6 + 0.12 * i && live; b.setAttribute("stroke", on ? INK : LIL); b.setAttribute("stroke-dasharray", on ? "none" : "6 7"); names[i].setAttribute("fill", on ? INK : MUTE); });+ const wkOn = t >= PR[3] + 0.1 && live;+ around(wk, VX + (3 * VW) / 4, FY[1], land(seg(t, PR[3] + 0.05, PR[3] + 0.35), 2.4).toFixed(3));+ wk.setAttribute("opacity", vis.toFixed(2));+ segT[1].setAttribute("fill", wkOn ? INK : MUTE);+ tf(st, VX + 8, FY[2], land(seg(t, PR[4] + 0.05, PR[4] + 0.4), 2.4).toFixed(3));+ st.setAttribute("opacity", vis.toFixed(2));+ [vL, ...vM].forEach((n) => n.setAttribute("opacity", vis.toFixed(2)));+ const tL = typed(vL, "morning check", t, 3.4, 4.4), t1 = typed(vM[0], "check something and", t, 6.6, 7.25), t2 = typed(vM[1], "write you three lines", t, 7.3, 7.95);+ const cn = tL ? vL : t1 ? vM[0] : t2 ? vM[1] : null;+ caret.setAttribute("opacity", cn && Math.sin(now * 9) > -0.3 ? 1 : 0);+ if (cn) { caret.setAttribute("x", (parseFloat(cn.getAttribute("x")) + cn.getComputedTextLength() + 3).toFixed(1)); caret.setAttribute("y", parseFloat(cn.getAttribute("y")) - 17); }+ /* the sun climbs, the card fires, the slip flies to the chair and is ticked */+ const rise = io(seg(t, 8.6, 10.0)) * (1 - io(seg(t, 13.0, 13.6))), a = ((212 - 74 * rise) * Math.PI) / 180;+ tf(sun, (AX + 104 * Math.cos(a)).toFixed(2), (HZ - 104 * Math.sin(a)).toFixed(2));+ rays.setAttribute("transform", `rotate(${((now * 14) % 360).toFixed(1)})`);+ const pulse = bump(t, 10.0);+ around(card, (X0 + X1) / 2, (Y0 + Y1) / 2, (1 + 0.014 * pulse).toFixed(4));+ const rk = seg(t, 10.0, 10.6);+ rip.setAttribute("cx", X1); rip.setAttribute("cy", 418);+ rip.setAttribute("r", (8 + 40 * oc(rk)).toFixed(1)); rip.setAttribute("opacity", (rk > 0 && rk < 1 ? 1 - rk : 0).toFixed(2));+ const fk = io(seg(t, 10.4, 11.2)), lw = t - 11.2;+ const sx = lerp(530, AX, fk), sy = lerp(418, FL - 81 - 24, fk) - 70 * Math.sin(Math.PI * fk);+ const rot = -12 * Math.sin(Math.PI * fk) + (lw > 0 ? 7 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0);+ tf(slip, sx.toFixed(2), sy.toFixed(2), land(seg(t, 10.05, 10.35), 2.2).toFixed(3), rot.toFixed(2));+ slip.setAttribute("opacity", t < 10.05 ? 0 : vis.toFixed(2));+ tf(ok, 16, -20, land(seg(t, 11.35, 11.7), 2.6).toFixed(3));+ const hop = 8 * Math.sin(Math.PI * seg(t, 6.0, 6.35)) + 6 * Math.sin(Math.PI * seg(t, 11.15, 11.45));+ tf(ag, AX, (FL - hop).toFixed(2));+ /* the pointer: fills the card, walks off, and peeks back in once to check */+ const [px, py] = at(t), wig = t > 11.3 && t < 11.9 ? 3 * Math.sin((t - 11.3) * 26) * Math.sin((Math.PI * (t - 11.3)) / 0.6) : 0;+ ptr.setAttribute("transform", `translate(${(px + 2 * press + wig).toFixed(2)} ${(py + 3 * press).toFixed(2)})`);+ say(cap, t < 0.9 || t >= 13.2 ? "Open Automations and press New automation" : t < 2.45 ? "Choose Schedule" : t < 4.9 ? "Then fill in four things" : t < 6.3 ? "When: an Interval or Weekly" : t < 8.0 ? "The message is what the agent should do when it fires" : t < 10.0 ? "A schedule sends a message to an agent on a clock" : "Every weekday morning, ask an agent to check something", 700);+ };+ },+ };++ /* ---- slack: switch Slack on under Connections and pick who answers; each card sends a note down its own string and the reply comes back into the same thread. Once, the strings go slack ---- */+ SC.slack = {+ dur: 14, still: 12.6,+ build(svg, api) {+ head(svg, "Send work from a channel", "tap a card");+ const CY = [176, 290, 404], XC = 628, FL = 474, S = 0.72, E = [XC - 46, FL - 80], M = [4.6, 5.4, 6.2], IN = 2.45;+ const NAMES = ["a channel", "another channel", "a direct message"];+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ /* Settings, then Connections: the Slack switch and Every channel the bot is in */+ el("rect", { x: 44, y: 84, width: 712, height: 72, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 66, 111, "Settings", { "font-size": 16, "font-weight": 600, fill: MUTE });+ txt(svg, 66, 139, "Connections", { "font-size": 22, "font-weight": 700 });+ txt(svg, 248, 128, "Slack", { "font-size": 20, "font-weight": 700 });+ const trk = el("rect", { x: 314, y: 106, width: 54, height: 28, rx: 14, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ const knob = el("circle", { cy: 120, r: 9, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const cb = el("rect", { x: 398, y: 107, width: 26, height: 26, rx: 7, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ const tick = el("path", { d: "M404.5 120.5 l5 5.5 l9 -11", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round", pathLength: 1, "stroke-dasharray": "0 1" }, svg);+ fit(txt(svg, 436, 128, "Every channel the bot is in", { "font-size": 20, "font-weight": 600 }), 300);+ /* who answers: an empty seat until the chair rolls in */+ fit(txt(svg, XC - 12, 292, "Who answers inbound messages", { "text-anchor": "middle", "font-size": 18, "font-weight": 700 }), 290);+ el("path", { d: `M440 ${FL + 2} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const ghost = el("g", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round", transform: `translate(${XC} ${FL}) scale(${S})` }, svg);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22 }, ghost);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11 }, ghost);+ el("path", { d: "M0 -80 V-14 M-38 -6 H38" }, ghost);+ /* one string from each card to the chair */+ const P0 = CY.map((y) => [372, y + 50]);+ const strs = CY.map(() => el("path", { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", pathLength: 1 }, svg));+ const ctl = (i, sag) => [(P0[i][0] + E[0]) / 2, (P0[i][1] + E[1]) / 2 + sag];+ const bez = (i, s, sag) => { const [x0, y0] = P0[i], [cx, cy] = ctl(i, sag), u = 1 - s; return [u * u * x0 + 2 * u * s * cx + s * s * E[0], u * u * y0 + 2 * u * s * cy + s * s * E[1]]; };+ /* the three chat cards: a channel, another channel, a direct message */+ const cards = CY.map((y0, i) => {+ const g = el("g", {}, svg);+ el("rect", { x: 44, y: y0, width: 328, height: 100, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("rect", { x: 37, y: y0 - 7, width: 342, height: 114, rx: 21, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, g);+ if (i < 2) el("path", { d: `M58 ${y0 + 15} h16 a5 5 0 0 1 5 5 v7 a5 5 0 0 1 -5 5 h-9 l-6 5 v-5 h-1 a5 5 0 0 1 -5 -5 v-7 a5 5 0 0 1 5 -5 z`, fill: SKY, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, g);+ else el("circle", { cx: 66, cy: y0 + 24, r: 9, fill: YEL, stroke: INK, "stroke-width": 2 }, g);+ txt(g, 88, y0 + 31, NAMES[i], { "font-size": 20, "font-weight": 700 });+ const msg = el("g", {}, g);+ el("rect", { x: 0, y: -11, width: 200, height: 22, rx: 11, fill: SOFT, stroke: INK, "stroke-width": 2 }, msg);+ el("path", { d: "M14 0 H122 M134 0 H180", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, msg);+ const rep = el("g", {}, g);+ el("path", { d: "M10 13 V22 Q10 30 18 30 H25", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, rep);+ const av = el("g", { transform: "translate(40 32) scale(0.74)" }, rep);+ el("rect", { x: -9, y: -17, width: 18, height: 14, rx: 5, fill: SKY, stroke: INK, "stroke-width": 2.4 }, av);+ el("rect", { x: -11, y: -2, width: 22, height: 6, rx: 3, fill: SKY, stroke: INK, "stroke-width": 2.4 }, av);+ el("path", { d: "M0 4 V11 M-8 13 H8", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, av);+ el("rect", { x: 56, y: 20, width: 200, height: 21, rx: 10.5, fill: MINT, stroke: INK, "stroke-width": 2 }, rep);+ el("path", { d: "M70 30.5 H170 M182 30.5 H232", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, rep);+ button(g, `Send a message from ${NAMES[i]}`, () => api.seek(M[i] - 0.15));+ return { msg, rep, y0 };+ });+ const ag = chair(svg, SKY, S);+ const note = CY.map(() => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -11, y: -14, width: 22, height: 28, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-5 -5 H5 M-5 2 H5 M-5 9 H1", stroke: MUTE, "stroke-width": 2.5, "stroke-linecap": "round" }, g);+ return g;+ });+ const back = CY.map(() => {+ const g = mv(el("g", { "pointer-events": "none" }, svg));+ el("rect", { x: -17, y: -10, width: 34, height: 20, rx: 10, fill: MINT, stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-8 0 H8", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ });+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ const PR = [0.6, 1.45, 2.3];+ const KP = [[0, 540, 232], [0.45, 354, 126], [0.95, 354, 126], [1.25, 412, 128], [1.75, 412, 128], [2.1, 652, 402], [2.45, 652, 402], [3.0, 880, 232], [13.4, 880, 232], [14, 540, 232]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ const bump = (t, a) => Math.sin(Math.PI * seg(t, a, a + 0.3));+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out, live = t < 13.2;+ /* the switch: on at the first press; once it drops out and comes back by itself */+ const g1 = io(seg(t, 10.4, 10.6)), g2 = land(seg(t, 11.3, 11.6), 2.4), on = io(seg(t, PR[0], PR[0] + 0.25)) * (1 - io(seg(t, 13.2, 13.6))) * (1 - g1 + g1 * g2);+ knob.setAttribute("cx", (328 + 26 * on).toFixed(2));+ trk.setAttribute("fill", on > 0.5 ? MINT : SOFT);+ const chk = t >= PR[1] + 0.05 && live;+ cb.setAttribute("fill", chk ? YEL : "#fff");+ tick.setAttribute("stroke-dasharray", `${(oc(seg(t, PR[1] + 0.05, PR[1] + 0.35)) * (live ? 1 : 0)).toFixed(3)} 1`);+ /* the chair rolls in to the empty seat */+ const k = seg(t, IN, IN + 1.0), x = lerp(880, XC, land(k, 1.15)), st = t - IN - 1.0;+ const lean = 3 * Math.sin(Math.PI * k) - (st > 0 ? 3 * Math.exp(-5 * st) * Math.sin(14 * st) : 0);+ const hop = M.reduce((h, m) => h + 6 * Math.sin(Math.PI * seg(t, m + 1.05, m + 1.35)), 0);+ tf(ag, x.toFixed(2), (FL - hop).toFixed(2), 1, lean.toFixed(2));+ ag.setAttribute("opacity", t < IN ? 0 : vis.toFixed(2));+ ghost.setAttribute("opacity", Math.max(1 - seg(t, IN + 0.6, IN + 1.0), out).toFixed(2));+ /* the strings: drawn out, taut; slack while Slack is off; a twang when it comes back */+ const tau = t - 11.3, twang = tau > 0 && tau < 1.2 ? 9 * Math.exp(-4 * tau) * Math.sin(22 * tau) : 0;+ const slack = 66 * io(seg(t, 10.4, 10.8)) * (1 - oc(seg(t, 11.3, 11.75))) + 3 * Math.sin(now * 3) * seg(t, 10.6, 10.9) * (1 - seg(t, 11.3, 11.5));+ const sag = 8 + slack + twang + (t > 12.2 && live ? 1.5 * Math.sin(now * 2) : 0);+ strs.forEach((p, i) => {+ const [cx, cy] = ctl(i, sag);+ p.setAttribute("d", `M${P0[i][0]} ${P0[i][1]} Q${cx.toFixed(1)} ${cy.toFixed(1)} ${E[0]} ${E[1]}`);+ p.setAttribute("stroke-dasharray", `${io(seg(t, 3.5 + 0.12 * i, 4.2 + 0.12 * i)).toFixed(3)} 1`);+ p.setAttribute("opacity", vis.toFixed(2));+ });+ /* each card: its message, the note down the string, the reply back, the reply in the thread */+ cards.forEach((c, i) => {+ const m = M[i];+ tf(c.msg, 66, c.y0 + 50, (land(seg(t, m, m + 0.35), 2.2) * vis).toFixed(3));+ tf(c.rep, 66, c.y0 + 50, (land(seg(t, m + 2.2, m + 2.55), 2.2) * vis).toFixed(3));+ const nk = io(seg(t, m + 0.25, m + 1.05)), [nx, ny] = bez(i, 0.07 + 0.89 * nk, sag);+ tf(note[i], nx.toFixed(2), ny.toFixed(2), land(seg(t, m + 0.25, m + 0.45), 2).toFixed(3), (8 * Math.sin(Math.PI * nk)).toFixed(2));+ note[i].setAttribute("opacity", t > m + 0.25 && t < m + 1.1 ? 1 : 0);+ const rk = io(seg(t, m + 1.4, m + 2.2)), [rx, ry] = bez(i, 0.96 - 0.88 * rk, sag);+ tf(back[i], rx.toFixed(2), ry.toFixed(2), land(seg(t, m + 1.4, m + 1.6), 2).toFixed(3), (-8 * Math.sin(Math.PI * rk)).toFixed(2));+ back[i].setAttribute("opacity", t > m + 1.4 && t < m + 2.22 ? 1 : 0);+ });+ const press = Math.max(...PR.map((a) => bump(t, a))), [px, py] = at(t);+ ptr.setAttribute("transform", `translate(${(px + 2 * press).toFixed(2)} ${(py + 3 * press).toFixed(2)})`);+ say(cap, t < 1.0 || t >= 13.2 ? "Under Settings, then Connections" : t < 2.0 ? "One bot can listen to Every channel the bot is in" : t < 3.6 ? "You choose the agent under Who answers inbound messages" : t < 6.2 ? "Send work from a channel and read the answer in the same thread" : t < 9.0 ? "A direct message to the bot is always answered" : t < 10.4 ? "The reply comes back in the same thread" : t < 12.0 ? "If Slack fails to start, it now comes back by itself" : "Read the answer in the same thread", 700);+ };+ },+ };++ /* ---- temp: a message comes in, a folding chair is opened (at the second try) beside the lasting chairs, takes the one job, reports done, folds flat and leaves ---- */+ SC.temp = {+ dur: 14, still: 9.9,+ build(svg, api) {+ head(svg, "A temp is an agent for one job", "tap the message");+ const FL = 444, XS = [124, 280, 436], COL = [SKY, LIL, MINT], TX = 640, FS = 1.15, HOME = [124, 140], WAIT = [612, 150], SEAT = [612, 318];+ const chair = (p, c, s) => {+ const g = el("g", { "pointer-events": "none" }, p), b = el("g", { transform: `scale(${s})` }, g);+ el("path", { d: "M-38 -10 L0 -24 L38 -10 M0 -24 V-10", fill: "none", stroke: INK, "stroke-width": 5, "stroke-linecap": "round", "stroke-linejoin": "round" }, b);+ [-38, 0, 38].forEach((x) => el("circle", { cx: x, cy: -6, r: 6, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b));+ el("rect", { x: -6, y: -86, width: 12, height: 64, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, b);+ el("path", { d: "M8 -90 H32", stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, b);+ el("circle", { cx: 35, cy: -90, r: 5, fill: YEL, stroke: INK, "stroke-width": 2 }, b);+ el("rect", { x: -7, y: -132, width: 14, height: 36, fill: INK }, b);+ el("path", { d: "M-62 -102 V-134 M62 -102 V-134", stroke: INK, "stroke-width": 4, "stroke-linecap": "round" }, b);+ el("path", { d: "M-74 -136 H-52 M52 -136 H74", stroke: INK, "stroke-width": 8, "stroke-linecap": "round" }, b);+ el("rect", { x: -50, y: -214, width: 100, height: 92, rx: 22, fill: c, stroke: INK, "stroke-width": 3 }, b);+ el("rect", { x: -56, y: -104, width: 112, height: 22, rx: 11, fill: c, stroke: INK, "stroke-width": 3 }, b);+ return g;+ };+ /* a folding chair seen from the side: back post, seat and front leg turn on one bolt */+ const folding = (p, ghost) => {+ const g = el("g", ghost ? { fill: "none", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 8", "stroke-linecap": "round" } : { "pointer-events": "none" }, p);+ const ln = ghost ? {} : { stroke: INK, "stroke-width": 7.5, "stroke-linecap": "round" }, sh = ghost ? {} : { fill: YEL, stroke: INK, "stroke-width": 3 };+ const leg = el("path", Object.assign({ d: "M24 -92 L-40 0" }, ln), g);+ el("path", Object.assign({ d: "M24 0 V-176" }, ln), g);+ const seat = el("rect", Object.assign({ x: -58, y: -92, width: 88, height: 18, rx: 8 }, sh), g);+ el("rect", Object.assign({ x: 6, y: -188, width: 38, height: 60, rx: 10 }, sh), g);+ if (!ghost) el("circle", { cx: 24, cy: -92, r: 4.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ return { g, leg, seat };+ };+ el("path", { d: `M44 ${FL + 2} H758`, stroke: INK, "stroke-width": 3, "stroke-linecap": "round" }, svg);+ const gh = folding(svg, true);+ tf(gh.g, TX, FL, FS);+ /* the lasting agents: the three office chairs from the team */+ const lc = XS.map((x, i) => chair(svg, COL[i], 0.95));+ el("path", { d: "M58 462 V472 H502 V462", fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, svg);+ txt(svg, 280, 500, "your lasting agents", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ /* the tag that hangs on the temp while it is here */+ const tag = el("g", {}, svg);+ const tp = pill(tag, "A temp", YEL), tw = parseFloat(tp.firstChild.getAttribute("width"));+ tf(tp, 716, 192);+ el("path", { d: `M${TX + 30} 232 Q${TX + 34} 206 ${716 - tw / 2 + 8} 196`, fill: "none", stroke: LIL, "stroke-width": 3, "stroke-linecap": "round" }, tag);+ const fw = el("g", {}, svg), fc = folding(fw, false);+ /* the slot the message waits in, which is also the tap target */+ const slot = el("g", {}, svg);+ el("rect", { x: HOME[0] - 72, y: HOME[1] - 36, width: 144, height: 72, rx: 18, fill: "#fff", stroke: LIL, "stroke-width": 2.5, "stroke-dasharray": "6 7" }, slot);+ el("rect", { x: HOME[0] - 80, y: HOME[1] - 44, width: 160, height: 92, rx: 22, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, slot);+ button(slot, "Send another message to a temp", () => api.seek(0.7));+ const dots = [0, 1, 2].map((j) => el("circle", { cx: SEAT[0] - 30 + 18 * j, cy: 270, r: 5.5, fill: LIL, stroke: INK, "stroke-width": 2 }, svg));+ const msg = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M-32 24 L-42 42 L-14 24 Z", fill: "#fff", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round" }, msg);+ el("rect", { x: -62, y: -28, width: 124, height: 56, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 3 }, msg);+ txt(msg, 0, 8, "one job", { "text-anchor": "middle", "font-size": 22, "font-weight": 700 });+ const done = mv(el("g", { "pointer-events": "none" }, svg));+ pill(done, "done", MINT, true);+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.2, 13.6), vis = 1 - out;+ /* the folding chair: slides in folded, half opens, snaps shut, opens properly; later folds and is dragged off */+ let u = 0;+ if (t >= 2.3 && t < 2.65) u = 0.5 * io(seg(t, 2.3, 2.65));+ else if (t >= 2.65 && t < 3.15) u = 0.5 - 0.42 * oc(seg(t, 2.65, 2.85));+ else if (t >= 3.15 && t < 8.1) u = 0.08 + 0.92 * land(seg(t, 3.15, 3.65), 1.4);+ else if (t >= 8.1 && t < 8.6) u = 1 - io(seg(t, 8.1, 8.5));+ const inK = seg(t, 1.1, 2.1), outK = io(seg(t, 9.0, 11.6));+ const fx = t < 8.6 ? lerp(900, TX, land(inK, 1.0)) : lerp(TX, 930, outK);+ const sh = t > 2.85 && t < 3.15 ? 1.6 * Math.sin((t - 2.85) * 50) * (1 - seg(t, 2.85, 3.15)) : 0;+ const lean = -8 * Math.sin(Math.PI * inK) + 12 * io(seg(t, 8.6, 9.0)) + sh;+ fc.leg.setAttribute("transform", `rotate(${(-34.8 * (1 - u)).toFixed(2)} 24 -92)`);+ fc.seat.setAttribute("transform", `rotate(${(90 * (1 - u)).toFixed(2)} 24 -92)`);+ tf(fc.g, fx.toFixed(2), (FL - 20 * FS * cl(1 - u)).toFixed(2), FS, lean.toFixed(2));+ gh.g.setAttribute("opacity", (1 - cl(u)).toFixed(2));+ /* marked as moving only while it slides in or out */+ if ((t > 1.1 && t < 2.3) || t > 8.6) fw.setAttribute("data-mv", ""); else fw.removeAttribute("data-mv");+ const tk = land(seg(t, 3.7, 4.0), 2.2) * (1 - seg(t, 7.9, 8.2));+ tag.setAttribute("opacity", tk > 0.02 ? Math.min(1, tk).toFixed(2) : 0);+ tf(tp, 716, 192, Math.min(1.2, tk).toFixed(3));+ /* the message: waits, flies over, hovers through the struggle, lands, is worked, and a new one arrives */+ let mx = HOME[0], my = HOME[1] + 3 * Math.sin(now * 2.2), ms = 1, mr = 0, mo = 1;+ if (t >= 0.8 && t < 1.0) ms = 1 + 0.06 * Math.sin(Math.PI * seg(t, 0.8, 1.0));+ else if (t >= 1.0 && t < 2.2) { const k = io(seg(t, 1.0, 2.2)); mx = lerp(HOME[0], WAIT[0], k); my = lerp(HOME[1], WAIT[1], k) - 22 * Math.sin(Math.PI * k); mr = 6 * Math.sin(Math.PI * k); }+ else if (t >= 2.2 && t < 3.85) { mx = WAIT[0]; my = WAIT[1] + 4 * Math.sin((t - 2.2) * 6); }+ else if (t >= 3.85 && t < 9.0) { const k = land(seg(t, 3.85, 4.35), 1.3); mx = lerp(WAIT[0], SEAT[0], cl(k)); my = lerp(WAIT[1], SEAT[1], k); ms = lerp(1, 0.8, cl(k)) * (1 - io(seg(t, 6.2, 6.5))); mo = t < 6.5 ? 1 : 0; }+ else if (t >= 9.0) { const k = land(seg(t, 9.0, 9.6), 1.4); mx = lerp(-90, HOME[0], k); mo = seg(t, 9.0, 9.12); }+ tf(msg, mx.toFixed(2), my.toFixed(2), ms.toFixed(3), mr.toFixed(2));+ msg.setAttribute("opacity", mo);+ dots.forEach((d, j) => { const w = t > 4.45 && t < 5.65 ? 1 : 0; d.setAttribute("opacity", w); d.setAttribute("cy", (270 - w * 6 * Math.max(0, Math.sin((t - 4.45) * 7 - j * 0.9))).toFixed(2)); });+ /* done pops up, flies to the first lasting chair, and that chair gives a small hop */+ const dk = io(seg(t, 6.7, 7.6)), lw = t - 7.6;+ const dx = lerp(SEAT[0] - 18, XS[0], dk), dy = lerp(222, 326, dk) - 90 * Math.sin(Math.PI * dk);+ tf(done, dx.toFixed(2), dy.toFixed(2), (land(seg(t, 5.7, 6.0), 2.4) * vis).toFixed(3), ((lw > 0 ? 6 * Math.exp(-5 * lw) * Math.sin(14 * lw) : 0) - 10 * Math.sin(Math.PI * dk)).toFixed(2));+ done.setAttribute("opacity", t < 5.7 ? 0 : vis.toFixed(2));+ lc.forEach((g, i) => tf(g, XS[i], (FL - (i === 0 ? 7 * Math.sin(Math.PI * seg(t, 7.6, 7.9)) : 0)).toFixed(2)));+ say(cap, t < 0.8 || t >= 13.2 ? "There is no button for one. You ask" : t < 3.8 ? "A temp can answer inbound messages" : t < 5.7 ? "It gets an objective, does the work" : t < 10.6 ? "Reports done and leaves" : "A temp answers one message or does one job", 700);+ };+ },+ };++ /* ---- tokens: slips fill a tall glass to the line and a press squashes them into one summary; on a switch of work the+ glass is turned over and shaken out (the summary clings on for one more shake), and a handoff.md note lands in it ---- */+ SC.tokens = {+ dur: 14, still: 12.6,+ build(svg, api) {+ head(svg, "Keep sessions small", "tap the glass");+ const GX = 214, GY = 356, PX = 360, PY = 196;+ /* the two rules, in Automations under Context, both switched on */+ el("rect", { x: 446, y: 92, width: 310, height: 290, rx: 18, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ txt(svg, 468, 126, "Automations", { "font-size": 17, "font-weight": 600, fill: MUTE });+ txt(svg, 468, 158, "Context", { "font-size": 26, "font-weight": 700 });+ const lab = [];+ const rows = [["Compact when", "context fills"], ["Clear on a switch", "of work"]].map((L, i) => {+ const y = 176 + i * 102, r = el("rect", { x: 462, y, width: 278, height: 90, rx: 14, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ L.forEach((s, j) => lab.push(fit(txt(svg, 480, y + 38 + j * 30, s, { "font-size": 21, "font-weight": 700 }), 184)));+ el("rect", { x: 676, y: y + 32, width: 50, height: 26, rx: 13, fill: MINT, stroke: INK, "stroke-width": 2.5 }, svg);+ el("circle", { cx: 713, cy: y + 45, r: 8.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ return r;+ });+ const fs = Math.min(...lab.map((n) => parseFloat(n.getAttribute("font-size"))));+ lab.forEach((n) => n.setAttribute("font-size", fs));+ const keep = el("g", {}, svg);+ pill(keep, "Leave them on", MINT, true);+ /* the coaster and the glass; the slips, the summary and the note live inside the glass and turn with it */+ el("rect", { x: GX - 110, y: 476, width: 220, height: 10, rx: 5, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ const G = mv(el("g", {}, svg));+ const OUT = "M-88 -110 L-76 108 Q-75 118 -65 118 H65 Q75 118 76 108 L88 -110";+ el("path", { d: OUT + " Z", fill: "#fff" }, G);+ const C = el("g", {}, G);+ el("path", { d: "M-76 100 H76 L76 108 Q75 118 65 118 H-65 Q-75 118 -76 108 Z", fill: SOFT }, G);+ el("path", { d: "M-80 -58 H80", stroke: LIL, "stroke-width": 3, "stroke-dasharray": "8 8", "stroke-linecap": "round" }, G);+ el("path", { d: "M-64 -88 L-58 40", stroke: SKY, "stroke-width": 7, "stroke-linecap": "round" }, G);+ el("path", { d: OUT, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linejoin": "round", "stroke-linecap": "round" }, G);+ el("path", { d: "M-76 100 H76", stroke: INK, "stroke-width": 2 }, G);+ const REST = [];+ for (let i = 0; i < 14; i++) REST.push([i % 2 ? 34 : -34, 89 - Math.floor(i / 2) * 22]);+ REST.push([-34, 53], [34, 53], [-28, 31]);+ const ROT = [-5, 4, 3, -4, -3, 5, 4, -3, -5, 2, 3, -4, -2, 4, 3, -4, 6];+ const TL = REST.map((_, i) => (i < 14 ? 0.7 + i * 0.22 : 6.8 + (i - 14) * 0.3));+ const S = REST.map(() => {+ const g = el("g", { "pointer-events": "none" }, C);+ el("rect", { x: -31, y: -9, width: 62, height: 18, rx: 4, fill: "#fff", stroke: INK, "stroke-width": 2 }, g);+ el("path", { d: "M-21 0 H13", stroke: MUTE, "stroke-width": 3, "stroke-linecap": "round" }, g);+ return g;+ });+ const sum = el("g", {}, C);+ el("rect", { x: -66, y: -17, width: 132, height: 34, rx: 6, fill: MINT, stroke: INK, "stroke-width": 2.5 }, sum);+ const sumT = txt(sum, 0, 6, "summary", { "text-anchor": "middle", "font-size": 18, "font-weight": 700 });+ const ho = el("g", {}, C);+ el("rect", { x: -66, y: -19, width: 132, height: 38, rx: 6, fill: YEL, stroke: INK, "stroke-width": 2.5 }, ho);+ txt(ho, 0, 6, "handoff.md", { "text-anchor": "middle", "font-size": 18, "font-weight": 700, "font-family": MONO });+ /* the press: a knob, a rod and a mesh disc, parked beside the glass */+ const P = mv(el("g", { "pointer-events": "none" }, svg));+ el("path", { d: "M0 -6 V-54", stroke: INK, "stroke-width": 6, "stroke-linecap": "round" }, P);+ el("circle", { cx: 0, cy: -66, r: 13, fill: YEL, stroke: INK, "stroke-width": 2.5 }, P);+ el("rect", { x: -70, y: -6, width: 140, height: 12, rx: 6, fill: SKY, stroke: INK, "stroke-width": 2.5 }, P);+ el("path", { d: "M-46 -6 V6 M-23 -6 V6 M0 -6 V6 M23 -6 V6 M46 -6 V6", stroke: INK, "stroke-width": 1.5 }, P);+ const hit = el("g", {}, svg);+ el("rect", { x: GX - 106, y: 230, width: 212, height: 266, rx: 18, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Fill the glass again", () => api.seek(0.3));+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the glass: turned over, shaken twice, turned back */+ const flip = io(seg(t, 8.0, 8.6)) + io(seg(t, 10.5, 11.1));+ const lift = 24 * (io(seg(t, 8.0, 8.4)) - io(seg(t, 10.7, 11.1)));+ const shA = t > 8.6 && t < 9.2 ? Math.sin((t - 8.6) * 46) * (1 - seg(t, 8.6, 9.2)) : 0;+ const shB = t > 9.5 && t < 10.1 ? Math.sin((t - 9.5) * 50) * (1 - seg(t, 9.5, 10.1)) : 0;+ const st = t - 11.1, settle = st > 0 && st < 0.7 ? 3 * Math.exp(-6 * st) * Math.sin(18 * st) : 0;+ const gr = 180 * flip + 5 * shA + 8 * shB + settle;+ tf(G, (GX + 7 * shA + 12 * shB).toFixed(2), (GY - lift).toFixed(2), 1, gr.toFixed(2));+ /* the press: parked with a small bob, swings over, presses down, comes back */+ const bob = (1 - seg(t, 3.8, 4.1) + seg(t, 6.8, 7.1)) * 2 * Math.sin(now * 2.2);+ let px = PX, py = PY + bob;+ if (t >= 4.1 && t < 4.8) { const k = io(seg(t, 4.1, 4.8)); px = lerp(PX, GX, k); py = lerp(PY, GY - 122, k) - 34 * Math.sin(Math.PI * k); }+ else if (t >= 4.8 && t < 5.65) { px = GX; py = GY + lerp(-122, 56, io(seg(t, 4.8, 5.4))); }+ else if (t >= 5.65 && t < 6.05) { px = GX; py = GY + lerp(56, -122, oc(seg(t, 5.65, 6.05))); }+ else if (t >= 6.05 && t < 6.8) { const k = seg(t, 6.05, 6.8); px = lerp(GX, PX, io(k)); py = lerp(GY - 122, PY, land(k, 1.2)) - 30 * Math.sin(Math.PI * io(k)); }+ tf(P, px.toFixed(2), py.toFixed(2));+ /* the slips: drop in, pile up, get squashed; the later ones fall out when the glass is turned over */+ const d = t >= 4.8 && t < 6.05 ? py - GY : -999, sq = cl((100 - (d + 6)) / 152, 0.12, 1), gone = seg(t, 5.42, 5.6);+ S.forEach((g, i) => {+ const [rx, ry] = REST[i], a = TL[i] - 0.45;+ let x = rx, y = ry, r = ROT[i], s = 1, sy = 1, o = t < a ? 0 : 1;+ if (t < TL[i]) { const k = seg(t, a, TL[i]); x = lerp(rx * 0.6, rx, io(k)); y = lerp(-196, ry, k * k); r = ROT[i] * (5 - 4 * k); s = Math.min(1.2, land(seg(t, a, a + 0.18), 2)); }+ else { const b = t - TL[i]; if (b < 0.3) y -= 6 * Math.sin((Math.PI * b) / 0.3) * (1 - b / 0.3); }+ if (i < 14) { y = 100 - (100 - y) * sq; sy = Math.min(1, Math.max(0.3, sq * 1.15)); o *= 1 - gone; }+ else if (t > 8.6) { const f0 = 8.62 + (i - 14) * 0.07, u = seg(t, f0, f0 + 0.45); y = lerp(ry, -180, u * u); r += 25 * u * (i % 2 ? 1 : -1); o *= 1 - seg(u, 0.55, 0.95); }+ g.setAttribute("transform", `translate(${x.toFixed(2)} ${y.toFixed(2)}) rotate(${r.toFixed(2)}) scale(${s.toFixed(3)} ${(s * sy).toFixed(3)})`);+ g.setAttribute("opacity", o.toFixed(2));+ g.setAttribute("display", o < 0.01 ? "none" : "inline");+ });+ /* the summary: pops out of the squash, clings through the first shake, drops on the second */+ let sY = 81, sR = 0, so = t < 5.4 ? 0 : 1;+ if (t > 8.6 && t < 9.75) sY = 81 + 1.5 * shA;+ if (t >= 9.75) { const u = seg(t, 9.75, 10.25); sY = lerp(81, -180, u * u); sR = -30 * u; so = 1 - seg(u, 0.55, 0.95); }+ tf(sum, 0, sY.toFixed(2), Math.min(1.3, land(seg(t, 5.4, 5.7), 2.2)).toFixed(3), sR.toFixed(2));+ /* the slip turns with the glass; its word flips over halfway, while it is on its side, so it never reads upside down */+ sumT.setAttribute("transform", Math.round(flip) === 1 ? "rotate(180 0 -0.5)" : "");+ sum.setAttribute("opacity", so.toFixed(2));+ sum.setAttribute("display", so < 0.01 ? "none" : "inline");+ /* the handoff note drops into the clean glass and stays */+ const hk = seg(t, 11.25, 11.7), hb = t - 11.7;+ const hy = t < 11.7 ? lerp(-196, 79, hk * hk) : 79 - (hb < 0.35 ? 7 * Math.sin((Math.PI * hb) / 0.35) * (1 - hb / 0.35) : 0);+ tf(ho, 0, hy.toFixed(2), Math.min(1.2, land(seg(t, 11.25, 11.45), 2)).toFixed(3), (t < 11.7 ? 8 * (1 - hk) : 0).toFixed(2));+ const hoo = t < 11.25 ? 0 : 1 - out;+ ho.setAttribute("opacity", hoo.toFixed(2));+ ho.setAttribute("display", hoo < 0.01 ? "none" : "inline");+ rows[0].setAttribute("fill", t >= 3.9 && t < 6.4 ? YEL : SOFT);+ rows[1].setAttribute("fill", t >= 7.7 && t < 11.9 ? YEL : SOFT);+ tf(keep, 601, 432, (1 + 0.12 * Math.sin(Math.PI * seg(t, 12.0, 12.35))).toFixed(3));+ say(cap, t < 0.5 || t >= 13.3 ? "Two defaults in 0.5.6 keep sessions small" : t < 3.9 ? "Long sessions fill an agent's context" : t < 6.4 ? "Compact when context fills summarises a long session" : t < 7.7 ? "and carries the summary forward" : t < 11.2 ? "Clear on a switch of work clears the session" : t < 12.0 ? "and leaves a handoff file to carry the thread" : "Both rules are in Automations under Context", 700);+ };+ },+ };++ /* ---- sidebar: the Pro view's six rows; a highlight walks down them and each row writes its one line in the pane.+ Once it is picked, the Stapler row floats a little above the sidebar ---- */+ SC.sidebar = {+ dur: 14, still: 12.4,+ build(svg, api) {+ head(svg, "One window with a sidebar", "tap a row");+ const X0 = 44, X1 = 756, Y0 = 88, YB = 140, Y1 = 498, SW = 262, NX = 304;+ const RY = [182, 238, 294, 350, 406, 462], LIT = [1.0, 2.7, 4.4, 6.1, 7.8, 9.5];+ const NAMES = ["Tasks", "Inbox", "Automations", "Memory", "Capabilities", "Stapler"];+ const NOTE = ["every card in the office", "one conversation per agent", "Scheduled, Context and Webhooks", "Search every agent's memory", "what each agent can reach", "a small window that floats above every app"];+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "#fff" }, svg);+ el("path", { d: `M${X0} ${YB} H${SW} V${Y1} H${X0 + 18} Q${X0} ${Y1} ${X0} ${Y1 - 18} Z`, fill: SOFT }, svg);+ [LIL, SKY, MINT].forEach((c, i) => el("circle", { cx: 72 + i * 24, cy: 114, r: 7, fill: c, stroke: INK, "stroke-width": 2 }, svg));+ /* the switch in the title bar */+ el("rect", { x: 560, y: 98, width: 180, height: 32, rx: 16, fill: SOFT, stroke: INK, "stroke-width": 2 }, svg);+ const thumb = el("rect", { x: 650, y: 101, width: 87, height: 26, rx: 13, fill: YEL, stroke: INK, "stroke-width": 2 }, svg);+ txt(svg, 605, 120, "Classic", { "text-anchor": "middle", "font-size": 17, "font-weight": 600, fill: MUTE });+ txt(svg, 693, 120, "PRO", { "text-anchor": "middle", "font-size": 17, "font-weight": 700 });+ /* the pane: a slot for each row's line, dashed while it is empty */+ const notes = NOTE.map((s, i) => {+ const bar = el("rect", { x: NX - 14, y: RY[i] - 21, height: 42, rx: 12, fill: SOFT, stroke: LIL, "stroke-width": 2, "stroke-dasharray": "6 6" }, svg);+ el("circle", { cx: 278, cy: RY[i], r: 5, fill: LIL, stroke: INK, "stroke-width": 2 }, svg);+ const n = fit(txt(svg, NX, RY[i] + 7, s, { "font-size": 21, "font-weight": 600 }), 420), w = n.getComputedTextLength();+ bar.setAttribute("width", (w + 28).toFixed(1));+ const id = `mg-sidebar-note-${i}`, cp = el("clipPath", { id }, svg);+ const cr = el("rect", { x: NX - 4, y: RY[i] - 24, width: 0, height: 48 }, cp);+ n.setAttribute("clip-path", `url(#${id})`);+ return { n, bar, cr, w };+ });+ /* the highlight, then the rows; the Stapler row's shadow shows when it floats */+ const shadow = el("ellipse", { cx: 153, cy: 487, rx: 78, ry: 3.5, fill: LIL }, svg);+ const hl = el("rect", { x: 54, y: -23, width: 198, height: 46, rx: 12, fill: YEL, stroke: INK, "stroke-width": 2 }, svg);+ const ICON = [+ (p, x, y) => { el("rect", { x: x - 11, y: y - 10, width: 22, height: 20, rx: 4, fill: SKY, stroke: INK, "stroke-width": 2 }, p); el("path", { d: `M${x - 4} ${y - 10} V${y + 10} M${x + 4} ${y - 10} V${y + 10}`, stroke: INK, "stroke-width": 2 }, p); },+ (p, x, y) => el("path", { d: `M${x - 10} ${y - 10} h20 a3 3 0 0 1 3 3 v10 a3 3 0 0 1 -3 3 h-11 l-7 6 v-6 h-2 a3 3 0 0 1 -3 -3 v-10 a3 3 0 0 1 3 -3 z`, fill: LIL, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, p),+ (p, x, y) => el("path", { d: `M${x + 3} ${y - 12} L${x - 8} ${y + 2} H${x} L${x - 3} ${y + 12} L${x + 8} ${y - 2} H${x} Z`, fill: YEL, stroke: INK, "stroke-width": 2, "stroke-linejoin": "round" }, p),+ (p, x, y) => { el("path", { d: `M${x - 8} ${y + 6} L${x} ${y - 7} L${x + 8} ${y + 6} Z`, fill: "none", stroke: INK, "stroke-width": 2 }, p); [[-8, 6], [0, -7], [8, 6]].forEach(([a, b]) => el("circle", { cx: x + a, cy: y + b, r: 4.5, fill: MINT, stroke: INK, "stroke-width": 2 }, p)); },+ (p, x, y) => star(p, x, y, 12, BLUE),+ (p, x, y) => { el("rect", { x: x - 13, y: y + 3, width: 26, height: 7, rx: 3.5, fill: "#fff", stroke: INK, "stroke-width": 2 }, p); el("rect", { x: x - 13, y: y - 8, width: 24, height: 8, rx: 4, fill: YEL, stroke: INK, "stroke-width": 2, transform: `rotate(-8 ${x + 10} ${y - 4})` }, p); el("circle", { cx: x + 10, cy: y + 1, r: 3, fill: INK }, p); },+ ];+ const rows = NAMES.map((s, i) => {+ const g = el("g", {}, svg), inner = el("g", {}, g);+ el("rect", { x: 50, y: RY[i] - 26, width: 206, height: 52, rx: 14, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, inner);+ ICON[i](inner, 80, RY[i]);+ fit(txt(inner, 104, RY[i] + 7, s, { "font-size": 21, "font-weight": 700 }), 142);+ button(g, `Show what ${s} is for`, () => api.seek(LIT[i] - 0.5));+ return inner;+ });+ /* the window's edge and lines go on last, so the border stays even */+ el("rect", { x: X0, y: Y0, width: X1 - X0, height: Y1 - Y0, rx: 18, fill: "none", stroke: INK, "stroke-width": 2.5, "pointer-events": "none" }, svg);+ el("path", { d: `M${X0} ${YB} H${X1} M${SW} ${YB} V${Y1}`, stroke: INK, "stroke-width": 2, "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the PRO side of the switch gives one small press at the start */+ const pr = 1 - 0.08 * Math.sin(Math.PI * seg(t, 0.2, 0.5));+ thumb.setAttribute("transform", `translate(693.5 114) scale(${pr.toFixed(3)}) translate(-693.5 -114)`);+ /* the Stapler row lifts off and floats once it is picked */+ const fl = io(seg(t, LIT[5], LIT[5] + 0.5)) * (1 - out), fy = -9 * fl + 2.2 * fl * Math.sin(now * 2.4);+ rows[5].setAttribute("transform", `translate(0 ${fy.toFixed(2)}) rotate(${(-1.5 * fl).toFixed(2)} 153 462)`);+ shadow.setAttribute("opacity", (0.85 * fl).toFixed(2));+ shadow.setAttribute("transform", `translate(153 0) scale(${(1 - 0.04 * Math.sin(now * 2.4) * fl).toFixed(3)} 1) translate(-153 0)`);+ /* the highlight walks down the rows, landing with a small overshoot */+ let hy = RY[0], on5 = 0;+ for (let j = 1; j < 6; j++) { const k = seg(t, LIT[j] - 0.45, LIT[j]); if (k > 0) hy = lerp(RY[j - 1], RY[j], land(k, 1.6)); if (j === 5) on5 = k; }+ const hs = Math.min(1.15, land(seg(t, 0.5, 0.85), 2)) * (1 - out);+ hl.setAttribute("transform", `translate(153 ${(hy + fy * on5).toFixed(2)}) rotate(${(-1.5 * fl).toFixed(2)}) scale(${hs.toFixed(3)}) translate(-153 0)`);+ hl.setAttribute("opacity", hs > 0.01 ? 1 : 0);+ /* each row writes its line; the current one keeps its soft slot */+ notes.forEach((nn, j) => {+ const k = oc(seg(t, LIT[j], LIT[j] + 0.6)) * (1 - out), cur = t >= LIT[j] && (j === 5 || t < LIT[j + 1]) && t < 13.3;+ nn.cr.setAttribute("width", ((nn.w + 8) * k).toFixed(1));+ nn.n.setAttribute("opacity", k > 0.01 ? 1 : 0);+ nn.bar.setAttribute("stroke-opacity", k > 0.4 ? 0 : 1);+ nn.bar.setAttribute("fill", cur || k < 0.4 ? SOFT : "#fff");+ });+ say(cap, t < 0.9 ? "The switch in the title bar moves between Classic and PRO" : "One window with a sidebar, and a screen for each part of the office", 700);+ };+ },+ };++ /* ---- askme: Ask me questions wait in a queue; the one in front gets a one line answer, its question mark turns+ into a tick, it moves on and the queue shuffles up (the last card is a little too keen and bumps the one ahead) ---- */+ SC.askme = {+ dur: 14, still: 11.0,+ build(svg, api) {+ head(svg, "Answering unblocks the task", "tap the card");+ const BX = 52, BY = 124, BW = 424, BH = 330, QX = 630, QY = [214, 326, 438], QW = 196, QH = 90, QB = [306, 280], FY = BY + 217;+ txt(svg, QX, 152, "waiting on a person", { "text-anchor": "middle", "font-size": 18, "font-weight": 600, fill: MUTE });+ const qcard = () => {+ const g = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: -QW / 2, y: -QH / 2, width: QW, height: QH, rx: 16, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ el("path", { d: "M-46 -14 H72 M-46 12 H40", stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.55 }, g);+ el("circle", { cx: -72, cy: -1, r: 15, fill: LIL, stroke: INK, "stroke-width": 2.5 }, g);+ txt(g, -72, 6, "?", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ return g;+ };+ const Q = [0, 1, 2, 3].map(qcard);+ /* the card that grows from the front of the queue into the big card */+ const mg = el("g", { "pointer-events": "none" }, svg);+ const mr = el("rect", { fill: "#fff", stroke: INK, "stroke-width": 2.5 }, mg);+ const ml = el("path", { stroke: MUTE, "stroke-linecap": "round", "stroke-opacity": 0.55 }, mg);+ const mb = el("g", {}, mg);+ el("circle", { r: 15, fill: LIL, stroke: INK, "stroke-width": 2.5 }, mb);+ txt(mb, 0, 7, "?", { "text-anchor": "middle", "font-size": 19, "font-weight": 700 });+ /* the big Ask me card */+ const BG = el("g", { "pointer-events": "none" }, svg);+ el("rect", { x: BX, y: BY, width: BW, height: BH, rx: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, BG);+ const chip = pill(BG, "Ask me", YEL);+ tf(chip, BX + 24 + parseFloat(chip.firstChild.getAttribute("width")) / 2, BY + 46);+ txt(BG, BX + BW - 26, BY + 53, "Inbox", { "text-anchor": "end", "font-size": 17, "font-weight": 600, fill: MUTE });+ el("path", { d: `M${BX + 30} ${BY + 100} H${BX + 372} M${BX + 30} ${BY + 128} H${BX + 342} M${BX + 30} ${BY + 156} H${BX + 220}`, stroke: MUTE, "stroke-width": 9, "stroke-linecap": "round", "stroke-opacity": 0.55 }, BG);+ const qb = el("g", {}, BG);+ el("circle", { r: 18, fill: LIL, stroke: INK, "stroke-width": 2.5 }, qb);+ txt(qb, 0, 8, "?", { "text-anchor": "middle", "font-size": 23, "font-weight": 700 });+ const tb = el("g", {}, BG);+ el("circle", { r: 18, fill: MINT, stroke: INK, "stroke-width": 2.5 }, tb);+ el("path", { d: "M-8 0 l5 6 l11 -12", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, tb);+ const field = el("rect", { x: BX + 24, y: FY - 27, width: BW - 48, height: 54, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 2 }, BG);+ const ph = txt(BG, BX + 44, FY + 6, "Answer in one line", { "font-size": 20, "font-weight": 500, fill: MUTE });+ const typed = el("path", { d: `M${BX + 46} ${FY} H${BX + 110} M${BX + 124} ${FY} H${BX + 186}`, stroke: INK, "stroke-width": 9, "stroke-linecap": "round", pathLength: 1, "stroke-dasharray": "0 1" }, BG);+ const caret = el("path", { d: `M0 ${FY - 15} V${FY + 15}`, stroke: BLUE, "stroke-width": 2.5, "stroke-linecap": "round" }, BG);+ const send = el("g", {}, BG);+ el("circle", { r: 18, fill: BLUE, stroke: INK, "stroke-width": 2.5 }, send);+ el("path", { d: "M-7 0 H6 M1 -6 L7 0 L1 6", fill: "none", stroke: "#fff", "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, send);+ const wait = pill(BG, "waiting on you", SKY);+ const ww = parseFloat(wait.firstChild.getAttribute("width"));+ const unb = pill(BG, "unblocks the task", MINT, true);+ const uw = parseFloat(unb.firstChild.getAttribute("width"));+ const hit = el("g", {}, svg);+ el("rect", { x: BX - 7, y: BY - 7, width: BW + 14, height: BH + 14, rx: 27, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, hit);+ button(hit, "Ask again", () => api.seek(0.3));+ const cap = strip(svg, 516, 740);+ return (t, now) => {+ /* two questions per loop: the first moves on at 3.3, the second holds until 12.6 */+ const L0 = t < 8 ? 3.3 : 12.6, M0 = L0 + 0.5;+ const lk = io(seg(t, L0, L0 + 0.5));+ const bo = t < L0 ? 1 : t < L0 + 0.5 ? 1 - lk : seg(t, L0 + 1.1, L0 + 1.25);+ const bx = t >= L0 && t < L0 + 0.5 ? -70 * lk : 0, br = t >= L0 && t < L0 + 0.5 ? -4 * lk : 0;+ BG.setAttribute("transform", `translate(${bx.toFixed(2)} 0) rotate(${br.toFixed(2)} ${BX + BW / 2} ${BY + BH / 2})`);+ BG.setAttribute("opacity", bo.toFixed(2));+ /* the answer: focus, type one short line, send */+ const ty = t < 3.8 ? seg(t, 0.9, 2.1) : t < 13.1 ? seg(t, 5.9, 7.1) : 0;+ const focus = (t > 0.7 && t < 2.3) || (t > 5.7 && t < 7.3);+ const tick = (t >= 2.5 && t < 3.8) || (t >= 7.5 && t < 13.1), tk = t < 5 ? seg(t, 2.5, 2.75) : seg(t, 7.5, 7.75);+ const press = Math.sin(Math.PI * seg(t, 2.25, 2.45)) + Math.sin(Math.PI * seg(t, 7.25, 7.45));+ field.setAttribute("stroke", focus ? BLUE : INK);+ ph.setAttribute("opacity", ty > 0 ? 0 : 1);+ typed.setAttribute("stroke-dasharray", `${ty.toFixed(3)} 1`);+ typed.setAttribute("opacity", ty > 0.005 ? 1 : 0);+ const len = ty * 126, cx = len <= 64 ? BX + 46 + len : BX + 124 + (len - 64);+ caret.setAttribute("transform", `translate(${(cx + (ty > 0 ? 9 : 0)).toFixed(2)} 0)`);+ caret.setAttribute("opacity", focus && Math.floor(now * 2.6) % 2 === 0 ? 1 : 0);+ tf(send, BX + BW - 50, FY, (1 - 0.14 * press).toFixed(3));+ const breathe = tick ? 0 : 0.05 * Math.sin(now * 3);+ tf(qb, QB[0], QB[1], (tick ? 1 - oc(tk) : 1 + breathe).toFixed(3));+ qb.setAttribute("opacity", tick && tk >= 1 ? 0 : 1);+ tf(tb, QB[0], QB[1], (tick ? Math.min(1.25, land(tk, 2.4)) : 0).toFixed(3));+ tb.setAttribute("opacity", tick ? 1 : 0);+ const wk = tick ? io(seg(tk, 0, 0.5)) : 0, uk = tick ? Math.min(1.2, land(seg(tk, 0.4, 1), 2.2)) : 0;+ tf(wait, BX + 24 + ww / 2, BY + 288, (1 - wk).toFixed(3));+ wait.setAttribute("opacity", wk >= 1 ? 0 : 1);+ tf(unb, BX + 24 + uw / 2, BY + 288, uk.toFixed(3));+ unb.setAttribute("opacity", uk > 0.01 ? 1 : 0);+ /* the big card grows out of the front of the queue */+ const mk = io(seg(t, M0, M0 + 0.65)), mvis = t >= M0 && t < M0 + 0.8;+ const x0 = QX - QW / 2, y0 = QY[0] - QH / 2;+ mr.setAttribute("x", lerp(x0, BX, mk).toFixed(2)); mr.setAttribute("y", lerp(y0, BY, mk).toFixed(2));+ mr.setAttribute("width", lerp(QW, BW, mk).toFixed(2)); mr.setAttribute("height", lerp(QH, BH, mk).toFixed(2));+ mr.setAttribute("rx", lerp(16, 22, mk).toFixed(2));+ const lx = lerp(x0 + 52, BX + 30, mk), l1 = lerp(x0 + 170, BX + 372, mk), l2 = lerp(x0 + 138, BX + 342, mk), ly1 = lerp(y0 + 31, BY + 100, mk), ly2 = lerp(y0 + 57, BY + 128, mk);+ ml.setAttribute("d", `M${lx.toFixed(1)} ${ly1.toFixed(1)} H${l1.toFixed(1)} M${lx.toFixed(1)} ${ly2.toFixed(1)} H${l2.toFixed(1)}`);+ ml.setAttribute("stroke-width", lerp(7, 9, mk).toFixed(2));+ tf(mb, lerp(QX - 72, QB[0], mk).toFixed(2), lerp(QY[0] - 1, QB[1], mk).toFixed(2), lerp(1, 1.2, mk).toFixed(3));+ mb.setAttribute("opacity", t < L0 + 1.1 ? 1 : 0);+ mg.setAttribute("opacity", mvis ? 1 : 0);+ /* the queue shuffles up; on the first shuffle the last card is too keen and bumps the one ahead */+ const pos = [[QX, QY[0], 0, 1], [QX, QY[1], 0, 1], [QX, QY[2], 0, 1], [QX + 80, QY[2], 0, 0]];+ const SH = t >= M0 && t < M0 + 1.4 ? [M0, L0 < 8 ? 1 : 0] : null;+ if (SH) {+ const [s0, bump] = SH, k1 = seg(t, s0, s0 + 0.5), k2 = seg(t, s0 + 0.15, s0 + 0.7), k3 = seg(t, s0 + 0.4, s0 + 0.9), hb = t - (s0 + 0.42);+ const nudge = bump && hb > 0 ? Math.exp(-5 * hb) * Math.sin(16 * hb) : 0;+ pos[0][3] = 0;+ pos[1] = [QX, lerp(QY[1], QY[0], land(k1, 1.3)) - 7 * nudge, 4 * nudge, 1];+ pos[2] = [QX, lerp(QY[2], QY[1], land(k2, bump ? 2.8 : 1.3)), -3 * nudge, 1];+ pos[3] = [lerp(QX + 80, QX, oc(k3)), QY[2], 0, cl(k3 * 3)];+ }+ Q.forEach((g, j) => { tf(g, pos[j][0].toFixed(2), pos[j][1].toFixed(2), 1, pos[j][2].toFixed(2)); g.setAttribute("opacity", pos[j][3].toFixed(2)); });+ say(cap, t < 0.7 ? "Ask me is the list of questions waiting on a person" : t < 2.3 ? "Answer in one line" : t < 3.3 ? "Answering unblocks the task" : t < 5.7 ? "and the card moves on" : t < 7.3 ? "Answer in one line" : t < 12.6 ? "A short answer unblocks the task, and the card moves on" : "This is the screen that keeps an office moving", 700);+ };+ },+ };++ /* ---- stapler: Show the stapler goes on, the round stapler pops up over a plain app, its ring opens, Screenshot takes+ a region, a note is added and stapled to it, and the pair goes to an agent; then the ring opens again ---- */+ SC.stapler = {+ dur: 14, still: 12.2,+ build(svg, api) {+ head(svg, "Staple it", "tap the stapler");+ const SX = 680, SY = 408, PCX = 335, PCY = 282, AX = 88, AY = 108;+ /* the setting */+ el("rect", { x: 448, y: 82, width: 308, height: 48, rx: 14, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ fit(txt(svg, 468, 113, "Show the stapler", { "font-size": 20, "font-weight": 700 }), 200);+ const trk = el("rect", { x: 684, y: 93, width: 52, height: 26, rx: 13, fill: SOFT, stroke: INK, "stroke-width": 2.5 }, svg);+ const knob = el("circle", { cy: 106, r: 8.5, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, svg);+ /* where captures go */+ const ag = el("g", {}, svg);+ const agR = el("rect", { x: 62, y: 88, height: 40, rx: 20, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, ag);+ const agT = txt(ag, 112, 115, "an agent", { "font-size": 20, "font-weight": 700 });+ agR.setAttribute("width", (agT.getComputedTextLength() + 72).toFixed(1));+ el("circle", { cx: AX, cy: AY, r: 12, fill: SKY, stroke: INK, "stroke-width": 2.5 }, ag);+ const agK = el("g", {}, ag);+ el("circle", { cx: AX, cy: AY, r: 12, fill: MINT, stroke: INK, "stroke-width": 2.5 }, agK);+ el("path", { d: `M${AX - 6} ${AY} l4 4.5 l8 -9`, fill: "none", stroke: INK, "stroke-width": 3, "stroke-linecap": "round", "stroke-linejoin": "round" }, agK);+ /* a plain app window with grey bars */+ el("rect", { x: 44, y: 146, width: 712, height: 346, rx: 16, fill: "#fff" }, svg);+ el("path", { d: "M44 182 H180 V492 H60 Q44 492 44 476 Z", fill: SOFT }, svg);+ [LIL, SKY, MINT].forEach((c, i) => el("circle", { cx: 68 + i * 22, cy: 164, r: 6.5, fill: c, stroke: INK, "stroke-width": 2 }, svg));+ el("rect", { x: 330, y: 158, width: 140, height: 12, rx: 6, fill: SOFT }, svg);+ const bars = (d) => el("path", { d, stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.45 }, svg);+ bars("M66 214 H150 M66 244 H136 M66 274 H154 M66 304 H128 M66 334 H146");+ [[202, 202, 264, 124], [486, 202, 250, 124], [202, 346, 534, 126]].forEach(([x, y, w, h]) => el("rect", { x, y, width: w, height: h, rx: 12, fill: SOFT }, svg));+ bars("M226 236 H400 M226 262 H440 M226 288 H350 M510 236 H690 M510 262 H650 M510 288 H700 M226 380 H600 M226 406 H520 M226 432 H640");+ el("rect", { x: 380, y: 276, width: 64, height: 38, rx: 8, fill: SKY }, svg);+ el("rect", { x: 44, y: 146, width: 712, height: 346, rx: 16, fill: "none", stroke: INK, "stroke-width": 2.5 }, svg);+ el("path", { d: "M44 182 H756", stroke: INK, "stroke-width": 2 }, svg);+ /* the region box drawn by the crosshair */+ const box = el("rect", { x: 212, y: 210, width: 0, height: 0, rx: 4, fill: SKY, "fill-opacity": 0.18, stroke: BLUE, "stroke-width": 2.5, "stroke-dasharray": "8 6" }, svg);+ /* the capture with its note, stapled; it flies as one */+ const P = mv(el("g", { "pointer-events": "none" }, svg)), Pin = el("g", { transform: `translate(${-PCX} ${-PCY})` }, P);+ el("rect", { x: 212, y: 210, width: 246, height: 108, rx: 10, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, Pin);+ el("path", { d: "M230 236 H400 M230 262 H436 M230 288 H350", stroke: MUTE, "stroke-width": 7, "stroke-linecap": "round", "stroke-opacity": 0.45 }, Pin);+ el("rect", { x: 380, y: 270, width: 60, height: 36, rx: 8, fill: SKY }, Pin);+ const note = el("g", {}, Pin);+ el("rect", { x: 212, y: 310, width: 246, height: 46, rx: 12, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, note);+ const nph = txt(note, 232, 340, "your note", { "font-size": 18, "font-weight": 500, fill: MUTE });+ const ntx = el("path", { d: "M234 334 H330 M344 334 H420", stroke: INK, "stroke-width": 7, "stroke-linecap": "round", pathLength: 1, "stroke-dasharray": "0 1" }, note);+ const staple = el("path", { d: "M-12 5 V-4 H12 V5", fill: "none", stroke: INK, "stroke-width": 3.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, Pin);+ /* the ring: four actions that open out of the stapler */+ const RING = [["Screenshot", 502, 420], ["Record message", 512, 344], ["Record meeting", 566, 274], ["Make invisible", 650, 216]];+ const ICO = [+ (g) => el("path", { d: "M-8 -3 V-8 H-3 M3 -8 H8 V-3 M8 3 V8 H3 M-3 8 H-8 V3", fill: "none", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "stroke-linejoin": "round" }, g),+ (g) => { el("rect", { x: -4, y: -10, width: 8, height: 13, rx: 4, fill: LIL, stroke: INK, "stroke-width": 2 }, g); el("path", { d: "M-7 0 Q-7 7 0 7 Q7 7 7 0 M0 7 V10", fill: "none", stroke: INK, "stroke-width": 2, "stroke-linecap": "round" }, g); },+ (g) => { el("circle", { r: 9.5, fill: "none", stroke: INK, "stroke-width": 2 }, g); el("circle", { r: 5.5, fill: BLUE }, g); },+ (g) => el("circle", { r: 8, fill: "none", stroke: MUTE, "stroke-width": 2.5, "stroke-dasharray": "3.5 3.5" }, g),+ ];+ const pills = RING.map(([s, x, y], i) => {+ const g = el("g", { "pointer-events": "none" }, svg);+ const r = el("rect", { y: -22, height: 44, rx: 22, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, g);+ const tx = txt(g, 0, 7, s, { "font-size": 20, "font-weight": 700 }), w = tx.getComputedTextLength() + 70;+ r.setAttribute("x", (-w / 2).toFixed(1)); r.setAttribute("width", w.toFixed(1)); tx.setAttribute("x", (-w / 2 + 50).toFixed(1));+ const ic = el("g", { transform: `translate(${(-w / 2 + 25).toFixed(1)} 0)` }, g);+ el("circle", { r: 16, fill: SOFT }, ic);+ ICO[i](ic);+ return { g, x, y, r };+ });+ /* the stapler: a dashed spot until the setting is on, then a round button that floats */+ const HG = el("g", {}, svg);+ const ghost = el("circle", { cx: SX, cy: SY, r: 34, fill: "#fff", "fill-opacity": 0, stroke: LIL, "stroke-width": 3, "stroke-dasharray": "7 7" }, HG);+ el("circle", { cx: SX, cy: SY, r: 42, fill: "#fff", "fill-opacity": 0, stroke: INK, "stroke-width": 2, class: "mg-ring" }, HG);+ button(HG, "Open the stapler ring", () => api.seek(1.45));+ const shadow = el("ellipse", { cx: SX, cy: SY + 50, rx: 26, ry: 5, fill: LIL, "pointer-events": "none" }, svg);+ const B = el("g", { "pointer-events": "none" }, svg);+ el("circle", { r: 34, fill: "#fff", stroke: INK, "stroke-width": 2.5 }, B);+ el("rect", { x: -23, y: 9, width: 46, height: 7, rx: 3.5, fill: SOFT, stroke: INK, "stroke-width": 2 }, B);+ el("rect", { x: 11, y: -5, width: 11, height: 17, rx: 3, fill: INK }, B);+ const arm = el("path", { d: "M18 -9 H-15 Q-23 -9 -23 -2 Q-23 3 -17 3 H18 Z", fill: YEL, stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round" }, B);+ const ptr = el("path", { d: "M0 0 V26 L7 19.5 L12 30 L16.5 28 L11.5 17.5 H21 Z", fill: "#fff", stroke: INK, "stroke-width": 2.5, "stroke-linejoin": "round", "pointer-events": "none" }, svg);+ const xh = el("path", { d: "M-13 0 H13 M0 -13 V13", stroke: INK, "stroke-width": 2.5, "stroke-linecap": "round", "pointer-events": "none" }, svg);+ const cap = strip(svg, 516, 740);+ const KP = [[0, 560, 300], [0.15, 560, 300], [0.45, 700, 110], [0.7, 700, 110], [1.35, 686, 412], [1.75, 686, 412], [2.3, 600, 470], [2.65, 444, 432], [3.1, 444, 432]];+ const at = (t) => { for (let i = 1; i < KP.length; i++) if (t < KP[i][0]) { const a = KP[i - 1], b = KP[i], k = io((t - a[0]) / (b[0] - a[0])); return [lerp(a[1], b[1], k), lerp(a[2], b[2], k)]; } return KP[KP.length - 1].slice(1); };+ return (t, now) => {+ const out = seg(t, 13.3, 13.7);+ /* the setting goes on, and the stapler pops up in its spot */+ const on = io(seg(t, 0.5, 0.75)) * (1 - io(seg(t, 13.3, 13.6)));+ knob.setAttribute("cx", (697 + 26 * on).toFixed(2));+ trk.setAttribute("fill", on > 0.5 ? MINT : SOFT);+ const pop = Math.min(1.2, land(seg(t, 0.75, 1.1), 2.2)) * (1 - io(seg(t, 13.4, 13.7)));+ const press = Math.sin(Math.PI * seg(t, 1.45, 1.65)), bobY = 3 * Math.sin(now * 2) * cl(pop);+ tf(B, SX, (SY + bobY).toFixed(2), (pop * (1 - 0.1 * press)).toFixed(3));+ B.setAttribute("opacity", pop > 0.01 ? 1 : 0);+ ghost.setAttribute("stroke-opacity", (1 - cl(pop)).toFixed(2));+ shadow.setAttribute("rx", (26 * cl(pop) * (1 - 0.06 * Math.sin(now * 2))).toFixed(2));+ shadow.setAttribute("opacity", (0.5 * cl(pop)).toFixed(2));+ /* the staple: the stapler's arm comes down, and a staple joins the note to the capture */+ const chomp = Math.sin(Math.PI * seg(t, 6.85, 7.1));+ arm.setAttribute("transform", `rotate(${(9 - 11 * chomp).toFixed(2)} 16 -3)`);+ /* the ring opens, Screenshot is picked, it closes; later it opens again and stays */+ pills.forEach((p, i) => {+ const o1 = land(seg(t, 1.65 + 0.07 * i, 2.05 + 0.07 * i), 1.6), c1 = i === 0 ? io(seg(t, 3.05, 3.3)) : io(seg(t, 2.9, 3.15));+ const o2 = land(seg(t, 9.3 + 0.08 * i, 9.75 + 0.08 * i), 1.6), c2 = io(seg(t, 13.2, 13.5));+ const k = t < 6 ? o1 * (1 - c1) : o2 * (1 - c2), pick = i === 0 ? Math.sin(Math.PI * seg(t, 2.8, 3.0)) : 0;+ tf(p.g, lerp(SX, p.x, cl(k)).toFixed(2), lerp(SY, p.y, cl(k)).toFixed(2), (Math.min(1.12, Math.max(0, k)) * (1 + 0.08 * pick)).toFixed(3));+ p.g.setAttribute("opacity", k > 0.02 ? 1 : 0);+ p.r.setAttribute("fill", i === 0 && t > 2.8 && t < 3.3 ? YEL : "#fff");+ });+ /* the pointer: the setting, the stapler, Screenshot; then the crosshair drags the region */+ const [px, py] = at(t), click = Math.max(Math.sin(Math.PI * seg(t, 0.5, 0.7)), press, Math.sin(Math.PI * seg(t, 2.8, 3.0)));+ ptr.setAttribute("transform", `translate(${(px + 2 * click).toFixed(2)} ${(py + 3 * click).toFixed(2)})`);+ ptr.setAttribute("opacity", t < 3.1 ? 1 : 0);+ const xk = io(seg(t, 3.1, 3.6)), dk = io(seg(t, 3.7, 4.7));+ const xx = t < 3.7 ? lerp(444, 212, xk) : lerp(212, 458, dk), xy = t < 3.7 ? lerp(432, 210, xk) : lerp(210, 318, dk);+ tf(xh, xx.toFixed(2), xy.toFixed(2));+ xh.setAttribute("opacity", t >= 3.1 && t < 4.9 ? 1 : 0);+ box.setAttribute("width", (246 * dk).toFixed(2)); box.setAttribute("height", (108 * dk).toFixed(2));+ box.setAttribute("opacity", t >= 3.7 && t < 5.0 ? (1 - seg(t, 4.85, 5.0)).toFixed(2) : 0);+ /* the capture lifts, the note slides in and is typed, the staple goes in, the pair flies to the agent */+ const lift = land(seg(t, 4.85, 5.15), 2), nk = oc(seg(t, 5.05, 5.45)), ty = seg(t, 5.7, 6.7);+ note.setAttribute("transform", `translate(0 ${(-30 * (1 - nk)).toFixed(2)})`);+ note.setAttribute("opacity", nk.toFixed(2));+ nph.setAttribute("opacity", ty > 0 ? 0 : 1);+ ntx.setAttribute("stroke-dasharray", `${ty.toFixed(3)} 1`);+ ntx.setAttribute("opacity", ty > 0.005 ? 1 : 0);+ const sk = seg(t, 6.95, 7.15);+ tf(staple, 234, 312, t < 6.95 ? 0 : Math.min(1.5, land(sk, 2.4)).toFixed(3));+ const fk = io(seg(t, 7.5, 8.4)), jig = t > 6.95 && t < 7.4 ? 2.5 * Math.sin((t - 6.95) * 40) * (1 - seg(t, 6.95, 7.4)) : 0;+ const fx = lerp(PCX, AX, fk), fy = lerp(PCY, AY, fk) + 24 * Math.sin(Math.PI * fk);+ tf(P, fx.toFixed(2), fy.toFixed(2), ((1 + 0.03 * (lift - cl(lift))) * (1 - 0.9 * fk)).toFixed(3), (jig - 8 * Math.sin(Math.PI * fk)).toFixed(2));+ const po = t < 4.85 || t > 8.4 ? 0 : 1 - seg(fk, 0.6, 0.85);+ P.setAttribute("opacity", po.toFixed(2));+ P.setAttribute("display", po < 0.01 ? "none" : "inline");+ /* the agent takes it, gives a small hop and a tick */+ const ak = seg(t, 8.35, 8.6), hop = 6 * Math.sin(Math.PI * seg(t, 8.35, 8.65));+ ag.setAttribute("transform", `translate(0 ${(-hop).toFixed(2)})`);+ tf(agK, 0, 0, 1);+ agK.setAttribute("opacity", t < 8.35 ? 0 : Math.min(1, ak * 3 * (1 - out)).toFixed(2));+ say(cap, t < 0.7 || t >= 13.3 ? "Turn on Show the stapler" : t < 1.55 ? "The Stapler is a small window that floats above every app" : t < 2.9 ? "Then click it for its ring" : t < 5.0 ? "Screenshot captures a region of the screen" : t < 6.9 ? "and sends it to an agent with a note" : t < 9.2 ? "When you see a bug in any app, do not describe it. Staple it" : "Its ring offers Screenshot, Record message, Record meeting and Make invisible", 700);+ };+ },+ };++ /* ==== ADD NEW SCENES ABOVE THIS LINE ==== */++ /* ---- player ---- */+ const q = new URLSearchParams(location.search), frozen = q.has("mgstill"), fixed = parseFloat(q.get("mgstill"));+ const calm = matchMedia("(prefers-reduced-motion: reduce)").matches;+ const mount = (fig) => {+ const sc = SC[fig.dataset.scene], img = fig.querySelector("img"); if (!sc || !img) return;+ const W = sc.w || 800, H = sc.h || 600;+ const svg = el("svg", { viewBox: `0 0 ${W} ${H}`, class: "mg-svg", role: "group", "aria-label": img.alt || "" });+ svg.style.aspectRatio = `${W} / ${H}`;+ el("rect", { width: W, height: H, fill: "#fff" }, svg);+ img.replaceWith(svg);+ let off = 0, raf = 0, seen = false, held = calm;+ const base = performance.now(), clock = () => (performance.now() - base) / 1000;+ let draw = () => {};+ const frame = () => { const now = clock(); draw(held ? sc.still : (((now + off) % sc.dur) + sc.dur) % sc.dur, now); };+ const tick = () => { cancelAnimationFrame(raf); frame(); if (seen && !held) raf = requestAnimationFrame(tick); };+ const api = { now: clock, seek: (s) => { off = s - clock(); held = false; tick(); }, poke: () => { held = false; tick(); } };+ draw = sc.build(svg, api);+ if (frozen) { draw(isNaN(fixed) ? sc.still : fixed, 0); return; }+ new IntersectionObserver(([e]) => {+ seen = e.isIntersecting;+ if (seen) { tick(); if (!fig.dataset.sent && window.posthog) { fig.dataset.sent = 1; window.posthog.capture("blog_scene_view", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } }+ else cancelAnimationFrame(raf);+ }, { threshold: 0.25 }).observe(fig);+ svg.addEventListener("click", () => { if (!fig.dataset.hit && window.posthog) { fig.dataset.hit = 1; window.posthog.capture("blog_scene_interact", { slug: location.pathname.split("/").filter(Boolean).pop(), scene: fig.dataset.scene }); } });+ frame();+ };+ const go = () => document.querySelectorAll("figure.mg[data-scene]").forEach(mount);+ const fonts = document.fonts && document.fonts.load ? Promise.all([document.fonts.load('700 32px "Space Grotesk"'), document.fonts.load('500 20px "Space Grotesk"'), document.fonts.load('700 20px "JetBrains Mono"')]).catch(() => {}) : Promise.resolve();+ const start = () => fonts.then(go, go);+ if (document.readyState === "loading") document.addEventListener("DOMContentLoaded", start); else start();+})();diff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.pngnew file mode 100644index 00000000..0107114eBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/schedule.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/schedule.pngnew file mode 100644index 00000000..434b5b08Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/schedule.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/sidebar.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/sidebar.pngnew file mode 100644index 00000000..ff83c9ffBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/sidebar.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/slack.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/slack.pngnew file mode 100644index 00000000..1f8ed85cBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/slack.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/stapler.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/stapler.pngnew file mode 100644index 00000000..017c71f0Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/stapler.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/team.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/team.pngnew file mode 100644index 00000000..ce41aa9dBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/team.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/temp.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/temp.pngnew file mode 100644index 00000000..f91cd899Binary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/temp.png differdiff --git a/docs/blog/assets/media/your-first-hour-with-munder-difflin/tokens.png b/docs/blog/assets/media/your-first-hour-with-munder-difflin/tokens.pngnew file mode 100644index 00000000..4cd452bfBinary files /dev/null and b/docs/blog/assets/media/your-first-hour-with-munder-difflin/tokens.png differdiff --git a/docs/blog/build-your-personal-agi-on-your-computer/index.html b/docs/blog/build-your-personal-agi-on-your-computer/index.htmlindex 7f3b5905..5e58569c 100644--- a/docs/blog/build-your-personal-agi-on-your-computer/index.html+++ b/docs/blog/build-your-personal-agi-on-your-computer/index.html@@ -342,50 +342,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/building-reliable-ai-agents/index.html b/docs/blog/building-reliable-ai-agents/index.htmlindex 42f7c8b2..fac31f6e 100644--- a/docs/blog/building-reliable-ai-agents/index.html+++ b/docs/blog/building-reliable-ai-agents/index.html@@ -358,50 +358,50 @@ by design; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/claude-code-automation-while-you-sleep/index.html b/docs/blog/claude-code-automation-while-you-sleep/index.htmlindex 7c785c5d..e81af345 100644--- a/docs/blog/claude-code-automation-while-you-sleep/index.html+++ b/docs/blog/claude-code-automation-while-you-sleep/index.html@@ -346,50 +346,50 @@ while you sleep; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/claude-code-max-plan-tips/index.html b/docs/blog/claude-code-max-plan-tips/index.htmlindex 90ef5409..b84359ce 100644--- a/docs/blog/claude-code-max-plan-tips/index.html+++ b/docs/blog/claude-code-max-plan-tips/index.html@@ -279,50 +279,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/claude-code-multi-agent-setup-tutorial/index.html b/docs/blog/claude-code-multi-agent-setup-tutorial/index.htmlindex 05e49dfd..ff5dfa34 100644--- a/docs/blog/claude-code-multi-agent-setup-tutorial/index.html+++ b/docs/blog/claude-code-multi-agent-setup-tutorial/index.html@@ -335,50 +335,50 @@ coordinated team of Claude Code agents — it’s free, local, and open source.<<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/claude-code-subagents-vs-multi-agent-harness/index.html b/docs/blog/claude-code-subagents-vs-multi-agent-harness/index.htmlindex 57636e8e..92ada4e5 100644--- a/docs/blog/claude-code-subagents-vs-multi-agent-harness/index.html+++ b/docs/blog/claude-code-subagents-vs-multi-agent-harness/index.html@@ -342,50 +342,50 @@ reports only failures. No Edit or Write tools, and use Haiku.</code></pre><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/codex-max-plan-tips/index.html b/docs/blog/codex-max-plan-tips/index.htmlindex 5849d533..13539381 100644--- a/docs/blog/codex-max-plan-tips/index.html+++ b/docs/blog/codex-max-plan-tips/index.html@@ -288,50 +288,50 @@ Usage: codex cloud exec [OPTIONS] --env <ENV_ID> [QUERY]<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/codex-pricing/index.html b/docs/blog/codex-pricing/index.htmlindex 0e34abc7..f0934954 100644--- a/docs/blog/codex-pricing/index.html+++ b/docs/blog/codex-pricing/index.html@@ -440,50 +440,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/command-center-guide/index.html b/docs/blog/command-center-guide/index.htmlindex 1242784c..6cb76163 100644--- a/docs/blog/command-center-guide/index.html+++ b/docs/blog/command-center-guide/index.html@@ -327,50 +327,50 @@ place in your morning routine, <a href="https://github.com/chaitanyagiri/munder-<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/connect-slack-to-munder-difflin/index.html b/docs/blog/connect-slack-to-munder-difflin/index.htmlindex 67e07096..1ec05d39 100644--- a/docs/blog/connect-slack-to-munder-difflin/index.html+++ b/docs/blog/connect-slack-to-munder-difflin/index.html@@ -334,50 +334,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/context-engineering-for-ai-agents/index.html b/docs/blog/context-engineering-for-ai-agents/index.htmlindex 707ecb33..811f5494 100644--- a/docs/blog/context-engineering-for-ai-agents/index.html+++ b/docs/blog/context-engineering-for-ai-agents/index.html@@ -364,50 +364,50 @@ it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/debugging-multi-agent-systems/index.html b/docs/blog/debugging-multi-agent-systems/index.htmlindex 59f17427..d1150848 100644--- a/docs/blog/debugging-multi-agent-systems/index.html+++ b/docs/blog/debugging-multi-agent-systems/index.html@@ -326,50 +326,50 @@ to run agents you can actually trace; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/index.html b/docs/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/index.htmlindex 0241f65e..ff728614 100644--- a/docs/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/index.html+++ b/docs/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/index.html@@ -322,50 +322,50 @@ rides your existing Claude subscription. Local-first, on your own plan, no surpr<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/evaluating-ai-agent-reliability/index.html b/docs/blog/evaluating-ai-agent-reliability/index.htmlindex 0741e275..541b99b7 100644--- a/docs/blog/evaluating-ai-agent-reliability/index.html+++ b/docs/blog/evaluating-ai-agent-reliability/index.html@@ -329,50 +329,50 @@ your bar.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/feed.xml b/docs/blog/feed.xmlindex 2a43612b..4326ff80 100644--- a/docs/blog/feed.xml+++ b/docs/blog/feed.xml@@ -4,9 +4,17 @@<subtitle>Guides, deep dives, and comparisons on running multi-agent Claude Code: orchestration, agent memory, automation, and the tooling landscape.</subtitle><link href="https://munderdiffl.in/blog/feed.xml" rel="self" /><link href="https://munderdiffl.in/blog/" />- <updated>2026-10-13T00:00:00.000Z</updated>+ <updated>2026-10-14T00:00:00.000Z</updated><id>https://munderdiffl.in/blog/</id><author><name>Chaitanya Giri</name><uri>https://munderdiffl.in</uri></author>+ <entry>+ <title>AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</title>+ <link href="https://munderdiffl.in/blog/ai-agent-use-cases-our-office/" />+ <id>https://munderdiffl.in/blog/ai-agent-use-cases-our-office/</id>+ <published>2026-10-14T00:00:00.000Z</published>+ <updated>2026-10-14T00:00:00.000Z</updated><category term="Guides" />+ <summary>We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</summary>+ </entry><entry><title>Agent Tools Today, 13 Oct 2026: Muse Wonder, GLM-Legal, the Codex Daemon Issue</title><link href="https://munderdiffl.in/blog/agent-tools-today-2026-10-13/" />@@ -479,12 +487,4 @@<updated>2026-09-15T00:00:00.000Z</updated><category term="Guides" /><summary>Turn Claude Code into a personal AGI on your own computer: agents with memory, schedules, Gmail, Calendar, voice and Slack, set up in Munder Difflin 0.5.2.</summary></entry>- <entry>- <title>How to Install Gemini CLI, and Can It Use Gemini 4 Yet?</title>- <link href="https://munderdiffl.in/blog/how-to-install-gemini-cli/" />- <id>https://munderdiffl.in/blog/how-to-install-gemini-cli/</id>- <published>2026-09-15T00:00:00.000Z</published>- <updated>2026-10-02T00:00:00.000Z</updated><category term="Guides" />- <summary>Install Gemini CLI 0.62.0 with one npm command, see which sign ins still work after 18 Jun 2026, and why Gemini 4 Argon isn't in it yet.</summary>- </entry></feed>diff --git a/docs/blog/gemini-4-argon-pricing/index.html b/docs/blog/gemini-4-argon-pricing/index.htmlindex eec271bb..fb4eaa99 100644--- a/docs/blog/gemini-4-argon-pricing/index.html+++ b/docs/blog/gemini-4-argon-pricing/index.html@@ -378,50 +378,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/codex-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/codex-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-05">Oct 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/codex-pricing/">Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)</a></h3>- <p class="dek">Codex pricing checked 5 Oct 2026: which ChatGPT plans include Codex, US prices from $0 to $500, how usage limits are counted, credits and API key rates.</p>+ <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>+ <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/codex-pricing/" aria-label="Read: Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)">Read →</a>+ <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a></div></div></article>diff --git a/docs/blog/generative-engine-optimization-techniques/index.html b/docs/blog/generative-engine-optimization-techniques/index.htmlindex 3e0044d6..e0fa5a40 100644--- a/docs/blog/generative-engine-optimization-techniques/index.html+++ b/docs/blog/generative-engine-optimization-techniques/index.html@@ -328,7 +328,7 @@ and quotable. <a href="/#install">Download Munder Difflin</a> to see it; free an</div><footer class="post-foot wrap"><div class="prevnext">- <a class="pn prev" href="/blog/how-to-install-and-use-munder-difflin/"><span class="k">← Older</span><span class="t">How to Install and Use Munder Difflin: A Beginner's Guide</span></a>+ <a class="pn prev" href="/blog/how-to-install-and-use-munder-difflin/"><span class="k">← Older</span><span class="t">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</span></a><a class="pn next" href="/blog/compressing-agent-memory/"><span class="k">Newer →</span><span class="t">Compressing Agent Memory Without Losing the Original</span></a></div><section class="related">diff --git a/docs/blog/how-to-add-a-github-copilot-cli-agent/index.html b/docs/blog/how-to-add-a-github-copilot-cli-agent/index.htmlindex 8edfb124..5f4cd473 100644--- a/docs/blog/how-to-add-a-github-copilot-cli-agent/index.html+++ b/docs/blog/how-to-add-a-github-copilot-cli-agent/index.html@@ -302,50 +302,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-add-an-mcp-server-to-claude-code/index.html b/docs/blog/how-to-add-an-mcp-server-to-claude-code/index.htmlindex f1e42826..b2f65b2c 100644--- a/docs/blog/how-to-add-an-mcp-server-to-claude-code/index.html+++ b/docs/blog/how-to-add-an-mcp-server-to-claude-code/index.html@@ -457,50 +457,50 @@ File modified: <span class="token operator"><</span>your folder<span class="t<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-connect-ollama-to-claude-code/index.html b/docs/blog/how-to-connect-ollama-to-claude-code/index.htmlindex 11957c59..f5df90bd 100644--- a/docs/blog/how-to-connect-ollama-to-claude-code/index.html+++ b/docs/blog/how-to-connect-ollama-to-claude-code/index.html@@ -298,50 +298,50 @@ API Error: Connection refused — a firewall or proxy may be blocking it (Connec<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-hire-from-the-agent-gallery/index.html b/docs/blog/how-to-hire-from-the-agent-gallery/index.htmlindex 52295dbb..3b102076 100644--- a/docs/blog/how-to-hire-from-the-agent-gallery/index.html+++ b/docs/blog/how-to-hire-from-the-agent-gallery/index.html@@ -313,50 +313,50 @@ gallery page, then import it with Add agent and import hire. Manifests are plain<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-install-and-use-munder-difflin/index.html b/docs/blog/how-to-install-and-use-munder-difflin/index.htmlindex 60a1f219..34bb5bf3 100644--- a/docs/blog/how-to-install-and-use-munder-difflin/index.html+++ b/docs/blog/how-to-install-and-use-munder-difflin/index.html@@ -18,8 +18,8 @@<meta charset="UTF-8" /><meta name="viewport" content="width=device-width, initial-scale=1.0" />-<title>How to Install and Use Munder Difflin: A Beginner's Guide — Munder Difflin Blog</title>-<meta name="description" content="A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux." />+<title>How to Install Munder Difflin 0.5.6: Beginner's Guide</title>+<meta name="description" content="Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure." /><link rel="canonical" href="https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/" /><link rel="icon" type="image/png" href="https://munderdiffl.in/logo.png" />@@ -29,13 +29,13 @@<!-- Open Graph --><meta property="og:site_name" content="Munder Difflin Blog" /><meta property="og:type" content="article" />-<meta property="og:title" content="How to Install and Use Munder Difflin: A Beginner's Guide — Munder Difflin Blog" />-<meta property="og:description" content="A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux." />+<meta property="og:title" content="How to Install Munder Difflin 0.5.6: Beginner's Guide" />+<meta property="og:description" content="Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure." /><meta property="og:url" content="https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/" /><meta property="og:image" content="https://munderdiffl.in/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png" /><meta property="og:image:width" content="1600" /><meta property="og:image:height" content="900" />-<meta property="og:image:alt" content="The first Munder Difflin onboarding screen, asking whether you are technical or non-technical" />+<meta property="og:image:alt" content="Title card on white. Large ink type reads Install Munder Difflin 0.5.6 under the line From zero to a working office, with a Beginner's guide pill, beside a shelf going up on two brackets with a spirit level on it, a screwdriver, a hex key and two spare screws." /><meta property="og:locale" content="en_US" /><meta property="article:published_time" content="2026-06-05T00:00:00.000Z" /><meta property="article:author" content="Chaitanya Giri" />@@ -43,15 +43,14 @@<meta property="article:tag" content="Getting Started" /><meta property="article:tag" content="Tutorial" /><meta property="article:tag" content="Non-Technical" />-<meta property="article:tag" content="Automation" /><!-- Twitter --><meta name="twitter:card" content="summary_large_image" />-<meta name="twitter:title" content="How to Install and Use Munder Difflin: A Beginner's Guide — Munder Difflin Blog" />-<meta name="twitter:description" content="A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux." />+<meta name="twitter:title" content="How to Install Munder Difflin 0.5.6: Beginner's Guide" />+<meta name="twitter:description" content="Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure." /><meta name="twitter:image" content="https://munderdiffl.in/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png" />-<meta name="twitter:image:alt" content="The first Munder Difflin onboarding screen, asking whether you are technical or non-technical" />+<meta name="twitter:image:alt" content="Title card on white. Large ink type reads Install Munder Difflin 0.5.6 under the line From zero to a working office, with a Beginner's guide pill, beside a shelf going up on two brackets with a spirit level on it, a screwdriver, a hex key and two spare screws." /><link rel="alternate" type="application/atom+xml" title="Munder Difflin Blog" href="https://munderdiffl.in/blog/feed.xml" /><link rel="preconnect" href="https://fonts.googleapis.com" />@@ -63,11 +62,11 @@{"@context": "https://schema.org","@type": "BlogPosting",- "headline": "How to Install and Use Munder Difflin: A Beginner's Guide",- "description": "A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.",+ "headline": "How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office",+ "description": "Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.","image": ["https://munderdiffl.in/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png"],"datePublished": "2026-06-05T00:00:00.000Z",- "dateModified": "2026-09-10T00:00:00.000Z",+ "dateModified": "2026-10-14T00:00:00.000Z","author": { "@type": "Person", "name": "Chaitanya Giri" },"publisher": {"@type": "Organization",@@ -75,7 +74,7 @@"logo": { "@type": "ImageObject", "url": "https://munderdiffl.in/logo.png" }},"mainEntityOfPage": { "@type": "WebPage", "@id": "https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/" },- "keywords": "how to install munder difflin, munder difflin download, munder difflin setup, what is a coding agent, munder difflin windows, free ai coding agent, antigravity free, munder difflin beginner guide",+ "keywords": "how to install munder difflin, munder difflin download, munder difflin setup, install munder difflin windows, install munder difflin mac, what is a coding agent, munder difflin beginner guide, munder difflin not installed fix","articleSection": "Guides","inLanguage": "en-US","url": "https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/"@@ -88,7 +87,7 @@"itemListElement": [{ "@type": "ListItem", "position": 1, "name": "Home", "item": "https://munderdiffl.in/" },{ "@type": "ListItem", "position": 2, "name": "Blog", "item": "https://munderdiffl.in/blog/" },- { "@type": "ListItem", "position": 3, "name": "How to Install and Use Munder Difflin: A Beginner's Guide", "item": "https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/" }+ { "@type": "ListItem", "position": 3, "name": "How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office", "item": "https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/" }]}</script>@@ -98,11 +97,12 @@"@context": "https://schema.org","@type": "FAQPage","mainEntity": [- { "@type": "Question", "name": "Do I need to know how to code to use Munder Difflin?", "acceptedAnswer": { "@type": "Answer", "text": "No. The onboarding asks whether you are technical or not on its very first screen, and the non-technical path replaces every piece of jargon with plain language. You will open a terminal exactly once, to install and log into your AI engine, and you can copy and paste those commands." } },- { "@type": "Question", "name": "Does Munder Difflin cost anything?", "acceptedAnswer": { "@type": "Answer", "text": "The classic app is free and open source, with unlimited local agents. Pro and Teams are optional paid plans. What you may pay for is the AI engine behind it. Antigravity is available at no charge and OpenCode has a free path, so you can run the whole thing without paying anything. If you already pay for ChatGPT or Claude, you can use those subscriptions instead." } },- { "@type": "Question", "name": "Can these agents really change files on my computer?", "acceptedAnswer": { "@type": "Answer", "text": "Yes, and that is the point, so it is worth understanding before you start. An agent can read files, write files and run commands in the folders you give it. You choose how much freedom it has during setup, and you can set it to ask permission before every change." } },- { "@type": "Question", "name": "Which operating systems does Munder Difflin run on?", "acceptedAnswer": { "@type": "Answer", "text": "macOS, Windows and Linux. macOS ships as one universal .dmg for Apple Silicon and Intel, Windows 10 and 11 as a setup installer or a portable build, and Linux as an AppImage." } },- { "@type": "Question", "name": "Is my code or data sent anywhere?", "acceptedAnswer": { "@type": "Answer", "text": "The app runs on your machine and stores its files there. Your prompts and the files an agent reads do go to whichever AI engine you picked, the same as if you used that tool directly. Anonymous usage telemetry is opt-out and every event is listed publicly in TELEMETRY.md." } }+ { "@type": "Question", "name": "Do I need to know how to code to install Munder Difflin?", "acceptedAnswer": { "@type": "Answer", "text": "No. You type two or three commands, to install an AI engine and sign in to it, and you can copy and paste them. If you skip that, the app offers to install a missing engine for you the first time an agent starts." } },+ { "@type": "Question", "name": "Do I have to install an AI engine before the app?", "acceptedAnswer": { "@type": "Answer", "text": "No. Setup marks each engine Installed, Installs on first start or Not installed. An engine marked Installs on first start is installed in the agent's own terminal when you press Install, and the app installs Node first if that engine needs it. You still have to sign in to the engine yourself." } },+ { "@type": "Question", "name": "Do I need an account?", "acceptedAnswer": { "@type": "Answer", "text": "Yes. The first launch asks you to sign in through your browser, on the free version as well. The browser does not open by itself: press Open browser to sign in and sign in there. The screen usually moves on by itself; if it does not, paste the code the browser shows you into the app." } },+ { "@type": "Question", "name": "Which operating systems does Munder Difflin run on?", "acceptedAnswer": { "@type": "Answer", "text": "macOS, Windows and Linux. macOS is one universal build for Apple Silicon and Intel, signed and notarized by Apple. Windows 10 and 11, 64 bit, get a setup installer and a portable build. Linux gets a 64 bit AppImage." } },+ { "@type": "Question", "name": "Can these agents really change files on my computer?", "acceptedAnswer": { "@type": "Answer", "text": "Yes, and that is the point, so understand it before you start. An agent starts in the folder you give it and can read files, write files and run commands. It is not locked in there: it can reach anything your account can. Start with a new empty folder and watch what the agents do before you point one at anything you care about." } },+ { "@type": "Question", "name": "Is the free version a separate download from Pro?", "acceptedAnswer": { "@type": "Answer", "text": "No. There is one installer. The free version is the Classic view, and a licence unlocks the Pro view in the same app. On first launch, press Continue with the free version to reach Classic." } }]}@@ -171,8 +171,8 @@<span>/</span><a href="/blog/topics/guides/">Guides</a></nav>- <h1>How to Install and Use Munder Difflin: A Beginner's Guide</h1>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h1>How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</h1>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="byline"><span class="av" aria-hidden="true">CG</span><strong>Chaitanya Giri</strong>@@ -184,404 +184,249 @@</div></header>- <div class="hero-wrap wrap"><figure class="fig fig-hero"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png" alt="The first Munder Difflin onboarding screen, asking whether you are technical or non-technical" loading="lazy" decoding="async" /></figure></div>+ <div class="hero-wrap wrap"><figure class="fig fig-hero"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png" alt="Title card on white. Large ink type reads Install Munder Difflin 0.5.6 under the line From zero to a working office, with a Beginner's guide pill, beside a shelf going up on two brackets with a spirit level on it, a screwdriver, a hex key and two spare screws." loading="lazy" decoding="async" /></figure></div><div class="post-wrap wrap"><div class="prose">- <div class="callout tldr"><span class="ic">TL;DR</span><p>Pick an AI engine based on what you-already pay for (or pick a free one). Install it with one command in the terminal, which is the-only time you will open a terminal. Download Munder Difflin, answer six onboarding screens, and-you have an office of AI workers you can watch on a screen.</p></div>-<p>If you have never used an AI coding tool, most guides start three steps ahead of you. This one does-not. It starts with what these things actually are, then walks the whole way to a working setup.</p>-<p>You will need about twenty minutes.</p>-<h2 id="first-what-is-munder-difflin" tabindex="-1">First, what is Munder Difflin? <a class="anchor" href="#first-what-is-munder-difflin" aria-hidden="true">#</a></h2>-<p>Munder Difflin is a desktop app that runs a small office of AI workers on your computer.</p>-<p>You see an actual office floor on screen, with characters at desks. Each character is an AI agent-working on something you asked for. One of them is your clone, the boss of the floor, who takes what-you ask for, breaks it into jobs, and hands those jobs to the others. You talk to your clone, and-your clone manages everyone else.</p>-<p>The app is free and open source. It runs on your machine, and it keeps its files on your machine. There are-optional paid plans, Pro and Teams, but nothing in this guide needs them.</p>-<p>The thing it is not: a chatbot. You are not sitting there typing and waiting for replies. You give-your clone a job, close the laptop, and come back to work that was done while you were gone.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/note-1.png" alt="Hand-drawn sketch from “How to Install and Use Munder Difflin”" loading="lazy" decoding="async" /></figure>-<h2 id="what-is-a-coding-agent-and-how-can-it-touch-my-computer" tabindex="-1">What is a coding agent, and how can it touch my computer? <a class="anchor" href="#what-is-a-coding-agent-and-how-can-it-touch-my-computer" aria-hidden="true">#</a></h2>-<p>This is the part worth reading slowly, because it is the part that surprises people.</p>-<p>A <strong>coding agent</strong> is an AI that does not just talk back. It can act. Give a normal chatbot a-question and it returns text. Give a coding agent a job and it will read your files, write new ones,-run commands, install things, and check its own work, in a loop, until the job is done or it gets-stuck.</p>-<p>It does this through a program on your computer called a <strong>CLI agent</strong>. CLI stands for command line-interface, which is the text-only way of controlling a computer that predates windows and buttons.-Claude Code, Codex, Antigravity and the others in this guide are all CLI agents. They are the engine.-Munder Difflin is the office that runs a team of them and keeps track of everything around them:-memory, tasks, schedules, files, and when to come ask you something.</p>-<p><strong>So how does it reach your computer?</strong> When you install one of these engines, it runs as a program-with your permissions. Anything you can do from your own account, it can technically do: open your-files, edit them, delete them, run software, reach the internet. It is not sandboxed away from your-machine by default. That is exactly why it is useful, and exactly why you should be deliberate.</p>-<p>Three things keep this sane, and Munder Difflin sets all three up during onboarding:</p>+ <div class="callout tldr"><span class="ic">TL;DR</span><p>Pick one AI engine, by what you already pay for. Download Munder Difflin 0.5.6 and open it. Press <strong>Open browser to sign in</strong> and sign in, name your workspace and your orchestrator, then press <strong>Continue with the free version</strong>. If the engine is missing, the app offers to install it. About twenty minutes.</p></div>+<p>This is how to install Munder Difflin 0.5.6 when you have never used an AI coding tool. Most guides start three steps ahead of you. This one starts with what the thing is, and ends with a working office and a fix for every way setup can fail.</p>+<h2 id="what-is-munder-difflin" tabindex="-1">What is Munder Difflin? <a class="anchor" href="#what-is-munder-difflin" aria-hidden="true">#</a></h2>+<p><a href="https://harnessmd.com/download">Munder Difflin</a> is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer.</p>+<p>In plain words, it is a small office of AI workers. One of them is your orchestrator, the one you talk to. He is called Michael until you rename him. You give Michael a job, he splits it into pieces, hands the pieces to other agents and comes back to you only when something needs your decision.</p>+<p>It is not a chatbot. You do not sit there typing and waiting for replies. You brief Michael, go and do something else, and come back to finished work.</p>+<figure class="mg" data-scene="office"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/office.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. An office seen from the front on white. One wide desk with a screen stands at the front, marked your orchestrator. Behind it, on a step marked other agents, are four smaller desks, each with its own screen. A slip marked your job lands on the front screen, is torn into three smaller slips, and each flies to a different desk at the back, where a progress bar starts to fill. Two of those desks finish with a tick. The third stops part way and sends back a slip marked needs your decision, which lands on the front screen. The fourth desk gets no slip, and its screen runs a bouncing screensaver. A filing cabinet and a water cooler stand at the sides."><figcaption>What Munder Difflin does: you brief one orchestrator, he hands the pieces to other agents, and only the decisions come back to you. Tap the front desk to send another job.</figcaption></figure>+<h2 id="what-is-a-coding-agent-and-what-can-it-touch-on-your-computer" tabindex="-1">What is a coding agent, and what can it touch on your computer? <a class="anchor" href="#what-is-a-coding-agent-and-what-can-it-touch-on-your-computer" aria-hidden="true">#</a></h2>+<p>A coding agent is an AI that acts instead of only answering. Ask a chatbot a question and you get text. Give a coding agent a job and it reads your files, writes new ones, runs commands and checks its own work, in a loop, until the job is done or it gets stuck.</p>+<p>Each agent runs through a program called an engine: Claude Code, Codex, Antigravity and the others in this guide. The engine does the thinking. Munder Difflin is the office around a team of them: tasks, memory, schedules, messages, and when to come and ask you.</p>+<p>An engine runs with your own permissions. Anything you can do from your account, it can do: open files, change them, delete them, run software. That is why it is useful, and why you should be deliberate. Three things keep it sane:</p><ol>-<li><strong>You choose which folders it can work in.</strong> Agents work inside the projects you add. A project-is just a folder. Do not point your first one at your entire home directory.</li>-<li><strong>You choose how much freedom it has.</strong> During setup you pick whether agents work on their own or-pause and ask you before changing files and running commands. If you are new, pick ask.</li>-<li><strong>You can watch it happen.</strong> Every agent has a terminal you can open and read, live. Nothing-happens off-screen.</li>+<li><strong>You choose the folder.</strong> An agent starts in the folder you give it. It is not locked in there, so do not make your first one your whole home folder.</li>+<li><strong>You choose how much freedom they have.</strong> Whether agents act on their own or stop and ask is a setting: <strong>Settings</strong>, then <strong>Autonomy & Budgets</strong>. A new install starts on autonomous: agents act without asking. Look at it before your first real job.</li>+<li><strong>You can watch.</strong> Every agent has a live terminal you can open and watch.</li></ol>-<div class="callout note"><span class="ic">Start safe</span><p>For your first project, make a brand-new empty folder and use that. Watch what the agents do for an hour before you point one at-anything you care about.</p></div>-<h2 id="step-1-which-ai-engine-will-you-use" tabindex="-1">Step 1: Which AI engine will you use? <a class="anchor" href="#step-1-which-ai-engine-will-you-use" aria-hidden="true">#</a></h2>-<p>Munder Difflin does not include an AI. It drives one you install. Twelve engines are supported, but-you only need one, and the right one is usually decided by what you already pay for.</p>-<p>Find yourself below.</p>-<h3 id="you-do-not-want-to-pay-anything" tabindex="-1">You do not want to pay anything <a class="anchor" href="#you-do-not-want-to-pay-anything" aria-hidden="true">#</a></h3>-<p><strong>Use Antigravity.</strong> It is Google’s agentic development platform and it is-<a href="https://antigravity.google/">available at no charge</a>. It ships with a command line tool called-<code>agy</code>, which is what Munder Difflin drives. This is the best starting point for most people reading-this guide.</p>-<p><strong>OpenCode</strong> is the other free path. It is open source, and it lets you bring your own model,-including free and local ones.</p>-<div class="callout note"><span class="ic">If you are in India, on Jio</span><p>Jio is running an-offer that gives eligible users <a href="https://www.jio.com/google-gemini-offer/">Google AI Pro-free for 18 months</a>. The terms on Jio's page: you must be over 18 and on an active unlimited 5G-plan of ₹349 or above, prepaid or postpaid, kept active for the whole period, claimable once per-number. Google AI Pro raises the daily request limits on <strong>Gemini CLI</strong>, which is a-separate supported engine in Munder Difflin. Note that Jio's page does not mention Antigravity, so-treat this as a boost to the Gemini CLI path rather than to Antigravity. Antigravity is free for-everyone anyway. Check Jio's page for current terms, offers change.</p></div>-<h3 id="you-already-pay-for-chatgpt" tabindex="-1">You already pay for ChatGPT <a class="anchor" href="#you-already-pay-for-chatgpt" aria-hidden="true">#</a></h3>-<p><strong>Use Codex.</strong> Your ChatGPT Plus or Pro subscription covers it. Nothing more to buy.</p>-<h3 id="you-already-pay-for-claude" tabindex="-1">You already pay for Claude <a class="anchor" href="#you-already-pay-for-claude" aria-hidden="true">#</a></h3>-<p><strong>Use Claude Code.</strong> Your Claude Pro or Max subscription covers it. This is the engine the app was-built against first, and it is the most thoroughly tested of the twelve.</p>-<h3 id="the-other-engines" tabindex="-1">The other engines <a class="anchor" href="#the-other-engines" aria-hidden="true">#</a></h3>-<p>Any of these work. Pick one only if you already use it.</p>+<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/autonomy-056.png" alt="Screenshot of Munder Difflin Settings, Autonomy & Budgets, in light mode: Autonomous set to autonomous, Who can add agents set to me and the orchestrator, the circuit breaker switched on with fields for a floor token budget and token velocity, and archiving settings." loading="lazy" decoding="async" /><figcaption>Settings, then Autonomy & Budgets: how much agents do alone, who may add agents, and the circuit breaker. From a preview build of 0.5.6 with sample data.</figcaption></figure>+<figure class="mg" data-scene="rooms"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/rooms.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. The floor plan of a house on white. Four pale rooms are each labelled as a folder: Documents, Photos, Projects and Downloads. Beside them is a hatched plot with a plus sign, labelled new empty folder. Its walls are drawn in and it becomes a fifth room, the only one drawn in colour. A small desk with a terminal on it drops into that room, a tag reading agent works here appears under it, and three new files appear beside it. The other rooms stay pale, and loose pages keep piling up in Downloads. Tapping a pale room colours it and moves a second desk in."><figcaption>An agent starts in the folder you give it. It is not locked in there: it can reach anything your account can. For your first job, make a new empty folder and use that. Tap a room to give the agent another folder.</figcaption></figure>+<h2 id="what-do-you-need-before-you-install" tabindex="-1">What do you need before you install? <a class="anchor" href="#what-do-you-need-before-you-install" aria-hidden="true">#</a></h2>+<p>Less than you would think. The app itself needs nothing else installed.</p>+<ul>+<li><strong>A computer.</strong> macOS on Apple Silicon or Intel, Windows 10 or 11 (64 bit), or a modern 64 bit Linux.</li>+<li><strong>An email address.</strong> The first launch asks you to sign in, on the free version as well.</li>+<li><strong>One AI engine, signed in.</strong> You can install it now with one command, or let the app install it later. Either way you sign in to the engine yourself, with your own subscription or key.</li>+</ul>+<p>Good to have, and you can add them after setup: git, and two small tools called <code>uv</code> and MemPalace that let agents search their memory by meaning. The app checks for these, and for Node, under <strong>Settings</strong>, then <strong>Prerequisites</strong>.</p>+<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/prerequisites-056.png" alt="Screenshot of Munder Difflin Settings, Prerequisites, in light mode: uv, git and Node.js marked Found with their paths, and MemPalace marked Not found with its install command and a copy button." loading="lazy" decoding="async" /><figcaption>Settings, then Prerequisites: each helper tool shows Found or Not found, and a missing one shows the command that installs it. From a preview build of 0.5.6 with sample data.</figcaption></figure>+<h2 id="which-ai-engine-should-you-pick" tabindex="-1">Which AI engine should you pick? <a class="anchor" href="#which-ai-engine-should-you-pick" aria-hidden="true">#</a></h2>+<p>Munder Difflin does not include an AI. It drives one you install. It supports twelve engines plus a custom one, and you need exactly one to start. What you already pay for usually decides it.</p>+<ul>+<li><strong>You pay for Claude:</strong> use <strong>Claude Code</strong>. It is the app’s default engine.</li>+<li><strong>You pay for ChatGPT:</strong> use <strong>Codex</strong>.</li>+<li><strong>You pay for nothing:</strong> use <strong>Antigravity</strong>, Google’s coding tool, whose command is <code>agy</code>.</li>+<li><strong>You already have a key from an AI provider:</strong> use <strong>OpenCode</strong>, which is open source and runs on the provider and model you choose.</li>+</ul>+<figure class="mg" data-scene="engine"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/engine.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three cards lie on a white table, reading You pay for Claude, You pay for ChatGPT and You pay for nothing. Each card turns over in turn to show the engine to pick: use Claude Code with a Claude subscription, use Codex with a ChatGPT subscription, and use Antigravity if you pay for nothing. The third card turns over once more than it needs to and lands crooked before it straightens. The line under the cards ends on You need exactly one to start."><figcaption>Pick your first engine by what you already pay for: Claude Code with a Claude subscription, Codex with a ChatGPT subscription, Antigravity if you pay for nothing. Tap a card to flip it back.</figcaption></figure>+<p>Each of the other engines has its own sign in, and the app can install them for you. Pick one only if you already use it.</p><div class="table-scroll"><table><thead><tr><th>Engine</th><th>Command</th>-<th>Good reason to pick it</th>+<th>Installs with</th></tr></thead><tbody><tr><td>Gemini CLI</td><td><code>gemini</code></td>-<td>You have Google AI Pro, including via the Jio offer</td>-</tr>-<tr>-<td>Grok</td>-<td><code>grok</code></td>-<td>You have an xAI subscription</td>-</tr>-<tr>-<td>Kimi Code</td>-<td><code>kimi</code></td>-<td>You use Kimi already</td>+<td>npm</td></tr><tr><td>Qwen</td><td><code>qwen</code></td>-<td>You want to run a model locally on your own hardware</td>+<td>npm</td></tr><tr><td>Crush</td><td><code>crush</code></td>-<td>You like Charm’s tools</td>+<td>npm</td>+</tr>+<tr>+<td>Pi</td>+<td><code>pi</code></td>+<td>npm</td></tr><tr><td>Copilot</td><td><code>copilot</code></td>-<td>You have GitHub Copilot</td>+<td>npm</td></tr><tr>-<td>Cursor</td>-<td><code>cursor-agent</code></td>-<td>You have a Cursor subscription</td>+<td>Grok</td>+<td><code>grok</code></td>+<td>its own installer</td></tr><tr>-<td>pi</td>-<td><code>pi</code></td>-<td>You use pi already</td>+<td>Kimi Code</td>+<td><code>kimi</code></td>+<td>its own installer</td>+</tr>+<tr>+<td>Cursor</td>+<td><code>cursor-agent</code></td>+<td>its own installer</td></tr></tbody></table></div>-<p>You can add more engines later and run several at once, with different agents on different engines.</p>-<h2 id="step-2-install-your-engine-and-log-in" tabindex="-1">Step 2: Install your engine, and log in <a class="anchor" href="#step-2-install-your-engine-and-log-in" aria-hidden="true">#</a></h2>-<p><strong>This is the only step where you open a terminal.</strong> Once it is done you will not need it again.</p>-<p>The terminal is a plain text window for typing commands. On macOS press <code>Cmd+Space</code>, type-<code>Terminal</code>, press Enter. On Windows press the Start button, type <code>PowerShell</code>, press Enter. On Linux-you already know.</p>-<p>Find your engine below, copy the block, paste it, press Enter. Wait for it to finish before typing-the next line.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-terminal.png" alt="A macOS Terminal window showing claude --version and codex --version both returning version numbers" loading="lazy" decoding="async" /><figcaption>Checking that the engines installed. This is the only time you need this window.</figcaption></figure>-<h3 id="first-do-you-have-nodejs" tabindex="-1">First: do you have Node.js? <a class="anchor" href="#first-do-you-have-nodejs" aria-hidden="true">#</a></h3>-<p>Most of these engines install through <code>npm</code>, which comes with Node.js. Check with:</p>-<pre class="language-bash"><code class="language-bash"><span class="token function">node</span> <span class="token parameter variable">--version</span></code></pre>-<p>If that prints a version number, you are ready. If it says “command not found”, get the <strong>LTS</strong>-build from <a href="https://nodejs.org/en/download">nodejs.org/en/download</a>, install it, then close the-terminal window and open a fresh one.</p>-<p>Antigravity, Claude Code and Cursor can be installed without Node.js. Everything else needs it.</p>-<h3 id="antigravity-free-no-subscription" tabindex="-1">Antigravity (free, no subscription) <a class="anchor" href="#antigravity-free-no-subscription" aria-hidden="true">#</a></h3>-<p>Google’s coding agent. The command is <code>agy</code>. Homepage:-<a href="https://antigravity.google/">antigravity.google</a> · Install docs:-<a href="https://antigravity.google/docs/cli/install/">antigravity.google/docs/cli/install</a></p>-<p>macOS and Linux:</p>+<p>You can add more engines later and run several at once, with different agents on different engines. The <a href="/blog/how-to-install-gemini-cli/">Gemini CLI install guide</a> has that engine’s steps.</p>+<h2 id="how-do-you-install-an-engine-and-sign-in" tabindex="-1">How do you install an engine and sign in? <a class="anchor" href="#how-do-you-install-an-engine-and-sign-in" aria-hidden="true">#</a></h2>+<p>Open a terminal and paste the lines for your engine one at a time, pressing Enter after each. A terminal is a plain text window for typing commands. On macOS press <code>Cmd+Space</code>, type <code>Terminal</code> and press Enter. On Windows press Start, type <code>PowerShell</code> and press Enter.</p>+<p><strong>Claude Code</strong> (a Claude subscription). macOS and Linux:</p>+<pre class="language-bash"><code class="language-bash"><span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://claude.ai/install.sh <span class="token operator">|</span> <span class="token function">bash</span></code></pre>+<p>Windows PowerShell:</p>+<pre class="language-powershell"><code class="language-powershell"><span class="token function">irm</span> https:<span class="token operator">/</span><span class="token operator">/</span>claude<span class="token punctuation">.</span>ai/install<span class="token punctuation">.</span>ps1 <span class="token punctuation">|</span> <span class="token function">iex</span></code></pre>+<p>On any system, close the terminal and open a new one, then run <code>claude</code> once and sign in when the browser opens.</p>+<p><strong>Codex</strong> (a ChatGPT subscription). macOS and Linux:</p>+<pre class="language-bash"><code class="language-bash"><span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://chatgpt.com/codex/install.sh <span class="token operator">|</span> <span class="token function">sh</span></code></pre>+<p>Windows PowerShell:</p>+<pre class="language-powershell"><code class="language-powershell">powershell <span class="token operator">-</span>ExecutionPolicy ByPass <span class="token operator">-</span>c <span class="token string">"irm https://chatgpt.com/codex/install.ps1 | iex"</span></code></pre>+<p>Codex can also install through npm, which comes with Node: <code>npm install -g @openai/codex</code>. Check Node with <code>node --version</code>; the app wants version 20 or newer. If you see an error such as “command not found” instead of a number, you have no Node: get the LTS build from <a href="https://nodejs.org/en/download">nodejs.org</a>, then close the terminal and open a new one.</p>+<p>Then close the terminal, open a new one and run <code>codex</code>. The first <code>codex</code> opens a sign in; choose to sign in with ChatGPT. <a href="/blog/how-to-install-codex-cli/">How to install Codex CLI</a> covers Windows and updates.</p>+<p><strong>Antigravity</strong> (no subscription). macOS and Linux:</p><pre class="language-bash"><code class="language-bash"><span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://antigravity.google/cli/install.sh <span class="token operator">|</span> <span class="token function">bash</span></code></pre><p>Windows PowerShell:</p><pre class="language-powershell"><code class="language-powershell"><span class="token function">irm</span> https:<span class="token operator">/</span><span class="token operator">/</span>antigravity<span class="token punctuation">.</span>google/<span class="token function">cli</span><span class="token operator">/</span>install<span class="token punctuation">.</span>ps1 <span class="token punctuation">|</span> <span class="token function">iex</span></code></pre>-<p>Then sign in:</p>-<pre class="language-bash"><code class="language-bash">agy</code></pre>-<p>Your browser opens. Sign in with a Google account.</p>-<h3 id="opencode-free-bring-your-own-model" tabindex="-1">OpenCode (free, bring your own model) <a class="anchor" href="#opencode-free-bring-your-own-model" aria-hidden="true">#</a></h3>-<p>Open source, works with free and local models. Docs:-<a href="https://opencode.ai/docs">opencode.ai/docs</a></p>+<p>Then close the terminal, open a new one, run <code>agy</code> and sign in with a Google account.</p>+<p><strong>OpenCode</strong> (a key from an AI provider you choose; you pay that provider). The second line asks which provider and takes the key:</p><pre class="language-bash"><code class="language-bash"><span class="token function">npm</span> <span class="token function">install</span> <span class="token parameter variable">-g</span> opencode-ai@latestopencode auth login</code></pre>-<h3 id="codex-with-a-chatgpt-subscription" tabindex="-1">Codex (with a ChatGPT subscription) <a class="anchor" href="#codex-with-a-chatgpt-subscription" aria-hidden="true">#</a></h3>-<p>Covered by ChatGPT Plus or Pro. Docs:-<a href="https://github.com/openai/codex">github.com/openai/codex</a></p>-<pre class="language-bash"><code class="language-bash"><span class="token function">npm</span> <span class="token function">install</span> <span class="token parameter variable">-g</span> @openai/codex-codex</code></pre>-<p>On first run it opens a browser to sign you into your ChatGPT account.</p>-<h3 id="claude-code-with-a-claude-subscription" tabindex="-1">Claude Code (with a Claude subscription) <a class="anchor" href="#claude-code-with-a-claude-subscription" aria-hidden="true">#</a></h3>-<p>Covered by Claude Pro or Max. Docs:-<a href="https://docs.claude.com/en/docs/claude-code">docs.claude.com/en/docs/claude-code</a></p>-<pre class="language-bash"><code class="language-bash"><span class="token function">npm</span> <span class="token function">install</span> <span class="token parameter variable">-g</span> @anthropic-ai/claude-code-claude</code></pre>-<p>If you would rather not install Node.js, use the standalone installer instead.</p>-<p>macOS and Linux:</p>-<pre class="language-bash"><code class="language-bash"><span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://claude.ai/install.sh <span class="token operator">|</span> <span class="token function">bash</span></code></pre>-<p>Windows PowerShell:</p>-<pre class="language-powershell"><code class="language-powershell"><span class="token function">irm</span> https:<span class="token operator">/</span><span class="token operator">/</span>claude<span class="token punctuation">.</span>ai/install<span class="token punctuation">.</span>ps1 <span class="token punctuation">|</span> <span class="token function">iex</span></code></pre>-<p>Then run <code>claude</code> and sign in when the browser opens.</p>-<h3 id="the-other-engines-1" tabindex="-1">The other engines <a class="anchor" href="#the-other-engines-1" aria-hidden="true">#</a></h3>-<p>Same idea: install, then run the command once on its own to sign in.</p>+<p>To check any of them, ask for its version, for example <code>claude --version</code>. A version number means you are done with the terminal.</p>+<figure class="mg" data-scene="terminal"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/terminal.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A light terminal window on white, above a row of keyboard keys with a large Enter key. A line types itself: npm install -g @openai/codex. Enter is pressed and a progress bar fills, slips back one block near the end, then finishes with a tick. A second line types codex, a dotted arrow runs from it to a small browser window that slides in over the terminal showing Sign in, and the button in that window turns into a tick. A third line types codex --version and prints three blocks that stand in for a version number, a tick and the words a version number."><figcaption>The whole terminal part of this guide: install the engine, run it once to sign in, then ask for its version to check. The drawing uses Codex; every engine follows the same three steps, though the install command differs. Tap the terminal to run it again.</figcaption></figure>+<div class="callout warn"><span class="ic">The step people skip</span><p><strong>Installing an engine is not the same as signing in to it.</strong> Run its command once and finish the sign in. Otherwise the agent cannot start. For Claude Code, Codex and Cursor the app shows <strong>Sign in needed</strong>.</p></div>+<h3 id="or-let-the-app-install-it" tabindex="-1">Or let the app install it <a class="anchor" href="#or-let-the-app-install-it" aria-hidden="true">#</a></h3>+<p>You can skip the terminal. Setup marks every engine <strong>Installed</strong>, <strong>Installs on first start</strong> or <strong>Not installed</strong>. When you start an agent whose engine is missing, the app says so and offers an <strong>Install</strong> button. Nothing installs until you press it.</p>+<p>The app uses the first route that can work on your computer: npm; Node and npm together if Node is missing; the engine’s own installer; else a manual route with a link to the docs. The sign in is still yours to finish, in that agent’s terminal.</p>+<h2 id="how-do-you-download-and-install-munder-difflin" tabindex="-1">How do you download and install Munder Difflin? <a class="anchor" href="#how-do-you-download-and-install-munder-difflin" aria-hidden="true">#</a></h2>+<p>Get the file for your system from the <a href="https://harnessmd.com/download">download page</a>.</p>+<figure class="mg" data-scene="download"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/download.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A download page on white shows the address harnessmd.com/download and three buttons labelled macOS, Windows and Linux. A file hangs from a clip under each button: universal .dmg, setup .exe with portable .exe beside it, and .AppImage. Each button is pressed in turn and lets go of its file, which opens a parachute and floats down to a landing pad on the ground. The portable .exe has a carry handle and no parachute, so it drops and bounces."><figcaption>One file per system: a universal .dmg for macOS, a setup .exe or a portable .exe for Windows, and an .AppImage for Linux. Tap a button to drop its file again.</figcaption></figure>+<ul>+<li><strong>macOS.</strong> One universal <code>.dmg</code> for Apple Silicon and Intel, signed and notarized by Apple. Open it and drag Munder Difflin into Applications. If macOS still refuses to open it, try to open it once, then open <strong>System Settings</strong>, then <strong>Privacy & Security</strong>, press <strong>Open Anyway</strong> within the hour and enter your password.</li>+<li><strong>Windows.</strong> Run the setup <code>.exe</code>. It installs for your user only and lets you pick the folder. There is also a portable <code>.exe</code> that needs no installing.</li>+<li><strong>Linux.</strong> Mark the <code>.AppImage</code> as executable (<code>chmod +x</code> on the file), then run it.</li>+</ul>+<p>Optional: to confirm the download arrived intact, compare its checksum with <code>SHA256SUMS-v0.5.6.txt</code>, published with the release. Run <code>shasum -a 256</code> on your download on macOS, <code>sha256sum</code> on Linux, or <code>Get-FileHash</code> in PowerShell, and compare the value with that file’s line.</p>+<h2 id="what-happens-the-first-time-you-open-it" tabindex="-1">What happens the first time you open it? <a class="anchor" href="#what-happens-the-first-time-you-open-it" aria-hidden="true">#</a></h2>+<p>Setup has four steps. The words in bold are the words on the screen.</p>+<ol>+<li><strong>How are you using Munder Difflin?</strong> Pick <strong>On my own</strong>, or <strong>With my team</strong> if someone sent you an invite code (a team code, not the sign in code below), and press <strong>Continue</strong>. Then comes <strong>Sign in to continue</strong>: press <strong>Open browser to sign in</strong>. The browser never opens by itself. Sign in there. The screen usually moves on by itself. If it does not, copy the code the browser shows, switch back to the app, paste it into <strong>Code from the browser</strong> and press <strong>Use this code</strong>. You see <strong>You are signed in</strong>; press <strong>Continue</strong>.</li>+<li><strong>Name your workspace.</strong> The default name is <code>HarnessAgents</code>. Everything your agents do lives in this one folder: their work, memory, tasks and history. <strong>Browse</strong> picks where it goes. The name cannot hold a space, and the app offers a fixed one if you type a space. This step also asks whether you are technical or non-technical.</li>+<li><strong>Meet your orchestrator.</strong> Name him and pick his engine. The engine you pick shows <strong>Installed</strong>, <strong>Installs on first start</strong> or <strong>Not installed</strong>.</li>+<li><strong>A last look at him.</strong> Press <strong>Open the workspace</strong>, and you are in.</li>+</ol>+<figure class="mg" data-scene="signin"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/signin.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two windows on white. An app window on the left holds a button reading Open browser to sign in, an empty field labelled Code from the browser and a pale button reading Use this code. A dotted outline on the right marks the browser, which is not open yet. The first button is pressed and a browser window opens in the outline, with a crowded row of tabs, a tick, the words copy the code and a code drawn as a row of boxes with a dot in each. A copy of the code lifts out of the browser, travels across and drops into the field. Use this code turns yellow and is pressed, and the app window shows You are signed in."><figcaption>Sign in takes two windows: press Open browser to sign in and sign in there. If the screen does not move on by itself, copy the code from the browser, paste it into Code from the browser and press Use this code. Tap the app window to watch it again.</figcaption></figure>+<p>Then comes the screen that stops people: the Pro sales screen, <strong>Run your agents like a company</strong>, with three buttons.</p>+<ul>+<li><strong>Continue with the free version</strong> opens the Classic view. This is the one you want if you have no licence.</li>+<li><strong>I have a licence key</strong> lets you paste a key.</li>+<li><strong>Get PRO</strong> opens the checkout in your browser.</li>+</ul>+<figure class="mg" data-scene="fork"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/fork.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A mini golf course on white, seen from above. A putter taps a ball off a mat marked after setup. The green lane splits into three branches, and a button sits on each one: Continue with the free version, I have a licence key and Get PRO. Each branch ends at a hole, marked Classic view, paste a key and checkout in your browser. The ball rolls to the split, stops, leans toward two of the branches, then takes the one under Continue with the free version, which turns yellow. It circles the rim of the hole once before it drops in, and a tick appears on the hole beside Classic view. The other two buttons then light up in turn. Tapping a branch sends the ball down that one."><figcaption>The first screen after setup offers Continue with the free version, I have a licence key and Get PRO. Without a licence, press Continue with the free version to reach the Classic view. Tap a branch to take it.</figcaption></figure>+<p>Classic is the free view: the office floor. The Pro view adds one window with screens for Tasks, Inbox, Automations, Memory and Capabilities, plus the Stapler, a small window above every app. You can move between the two later with the switch in the title bar, <strong>Classic</strong> and <strong>PRO</strong>. Without a licence the second one reads <strong>TRY PRO</strong>, and pressing it brings back this screen.</p>+<h2 id="what-do-you-do-in-your-first-ten-minutes" tabindex="-1">What do you do in your first ten minutes? <a class="anchor" href="#what-do-you-do-in-your-first-ten-minutes" aria-hidden="true">#</a></h2>+<p>You are in the office. Your orchestrator is there, and nobody else yet. Talk to him the way you would brief a capable new hire: the outcome, the folder, and what done looks like.</p>+<p>Weak: <em>“organise my files”</em></p>+<p>Better: <em>“Look in my Documents/receipts folder. Sort every PDF into folders by year, using the date inside the document, not the file name. Tell me about any file you cannot read.”</em></p>+<p>Your orchestrator decides whether he needs help, and hires it. A new agent appears at a desk. When you hire one yourself, you pick its face from fifteen built in avatars. Click any agent to open its terminal and watch it work.</p>+<p>When that works, <a href="/blog/your-first-hour-with-munder-difflin/">your first hour</a> covers teams, schedules, Slack and the Pro screens. For ideas, <a href="/blog/ai-agent-use-cases-our-office/">every agent on our own floor</a> shows the jobs we give ours.</p>+<h2 id="what-is-the-ide-button-for" tabindex="-1">What is the IDE button for? <a class="anchor" href="#what-is-the-ide-button-for" aria-hidden="true">#</a></h2>+<p>Click an agent and press <strong>IDE</strong> in its panel. The button opens a code editor on that agent’s working folder, inside the app. Use it to see what an agent just did without leaving the office.</p>+<ul>+<li>Open, edit and save any file, or make a <strong>New file</strong>.</li>+<li>See what the agent changed: <strong>Changes</strong>, <strong>History</strong> and <strong>Compare</strong>.</li>+<li>Preview Markdown files, the plain text format agents write reports in, as <strong>code</strong>, <strong>split</strong> or <strong>preview</strong>.</li>+<li>Open images.</li>+</ul>+<p>It is the quickest way to check an agent’s work. Ask for a report, open the IDE and read the report in preview next to the files it describes.</p>+<figure class="mg" data-scene="ide"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/ide.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A light code editor window on white. Its title bar holds a button reading IDE and the words an agent's working folder. Under it are three tabs, Changes, History and Compare, and a switch reading code, split and preview. A file list runs down the left side with a New file button at the bottom, and one file is selected. The code pane shows the file as grey bars, one bar per line. Two new lines type themselves in at the bottom, and a yellow dot marks the selected file as changed. The Changes tab lights up and the two new lines are marked in green. The switch then slides to split, and a formatted preview with a heading and a bullet list opens beside the code. It slides to preview, where the preview fills the pane, and back to split. The scroll bar handle is tiny, because the file is long."><figcaption>The IDE button opens an editor on an agent's working folder: edit and save files, see Changes, History and Compare, and preview Markdown as code, split or preview. Tap the switch to change the view.</figcaption></figure>+<h2 id="what-can-you-add-after-setup" tabindex="-1">What can you add after setup? <a class="anchor" href="#what-can-you-add-after-setup" aria-hidden="true">#</a></h2>+<p>Almost everything. Setup asks for the minimum, and the rest lives in <strong>Settings</strong>.</p>+<ul>+<li><strong>Prerequisites</strong> shows whether each engine and helper tool is installed.</li>+<li><strong>Agents & Models</strong> is where a second or third engine comes in.</li>+<li><strong>Autonomy & Budgets</strong> holds how much agents may do alone, their budgets, and whether temps (short lived agents for one job) may be started.</li>+<li><strong>Connections</strong> is Slack and the other ways work arrives from outside. The <a href="/blog/connect-slack-to-munder-difflin/">Slack guide</a> walks through it.</li>+<li><strong>Memory & Knowledge</strong> covers what agents remember. Search by meaning needs <code>uv</code> and MemPalace.</li>+<li><strong>Keys & Secrets</strong> is where saved keys live. OpenCode, Crush, Pi and Qwen take a key; Claude Code and Codex use their own sign in.</li>+</ul>+<h2 id="what-do-you-do-when-an-install-fails" tabindex="-1">What do you do when an install fails? <a class="anchor" href="#what-do-you-do-when-an-install-fails" aria-hidden="true">#</a></h2>+<p>Read the message first. The app says what is wrong in a sentence, and most screens carry the button that fixes it.</p><div class="table-scroll"><table><thead><tr>-<th>Engine</th>-<th>Install it with</th>-<th>Then run</th>-<th>Official docs</th>+<th>What you see</th>+<th>What to do</th></tr></thead><tbody><tr>-<td>Gemini CLI</td>-<td><code>npm install -g @google/gemini-cli</code></td>-<td><code>gemini</code></td>-<td><a href="https://github.com/google-gemini/gemini-cli">google-gemini/gemini-cli</a></td>-</tr>-<tr>-<td>Grok</td>-<td><code>npm install -g @xai-official/grok</code></td>-<td><code>grok</code></td>-<td><a href="https://docs.x.ai/build/overview">docs.x.ai</a></td>+<td>“… is not installed on this computer.”, with your engine’s name in front</td>+<td>Press <strong>Install</strong>. If it stops, <strong>Try again</strong>. <strong>Open the docs</strong> and <strong>Set up manually</strong> are the fallbacks.</td></tr><tr>-<td>Kimi Code</td>-<td><code>npm install -g @moonshot-ai/kimi-code</code></td>-<td><code>kimi</code></td>-<td><a href="https://www.kimi.com/code/docs/en/kimi-code-cli/guides/getting-started.html">kimi.com/code</a></td>+<td><strong>Sign in needed</strong></td>+<td>Finish the engine’s sign in in that agent’s terminal, then press <strong>Setup complete, start agent</strong>.</td></tr><tr>-<td>Qwen</td>-<td><code>npm install -g @qwen-code/qwen-code@latest</code></td>-<td><code>qwen</code></td>-<td><a href="https://github.com/QwenLM/qwen-code">QwenLM/qwen-code</a></td>+<td>“Signing in needs the internet. Your work is safe on this machine. Reconnect and try again.”</td>+<td>Reconnect, then try again. Nothing is lost.</td></tr><tr>-<td>Crush</td>-<td><code>npm install -g @charmland/crush</code></td>-<td><code>crush</code></td>-<td><a href="https://github.com/charmbracelet/crush">charmbracelet/crush</a></td>+<td>“Munder Difflin could not verify the secure connection to our server.”</td>+<td>Check the computer’s date and time. If they are right, ask whoever runs your network or security software to allow the address in the message, then press <strong>Try again</strong>.</td></tr><tr>-<td>Copilot</td>-<td><code>npm install -g @github/copilot</code></td>-<td><code>copilot</code></td>-<td><a href="https://docs.github.com/copilot/concepts/agents/about-copilot-cli">docs.github.com</a></td>+<td>“A folder name cannot hold a space.”</td>+<td>Press the button that reads <strong>Use</strong> and the suggested name.</td></tr><tr>-<td>Cursor</td>-<td><code>curl https://cursor.com/install -fsS | bash</code></td>-<td><code>cursor-agent</code></td>-<td><a href="https://cursor.com/docs/cli/overview">cursor.com/docs</a></td>+<td>A note that a port (a numbered door programs listen on) was in use and the session listens on another one</td>+<td>Nothing, unless something on this computer calls that port, such as a webhook. Then point it at the new number.</td></tr><tr>-<td>pi</td>-<td><code>npm install -g --ignore-scripts @earendil-works/pi-coding-agent</code></td>-<td><code>pi</code></td>-<td><a href="https://pi.dev/docs/latest">pi.dev/docs</a></td>+<td>macOS will not open the app</td>+<td>Try to open it once, then <strong>System Settings</strong>, then <strong>Privacy & Security</strong>, then <strong>Open Anyway</strong>.</td></tr></tbody></table></div>-<p>On Windows, the Cursor line is <code>irm 'https://cursor.com/install?win32=true' | iex</code> in PowerShell.</p>-<h3 id="check-it-actually-installed" tabindex="-1">Check it actually installed <a class="anchor" href="#check-it-actually-installed" aria-hidden="true">#</a></h3>-<p>Run this, swapping <code>claude</code> for whichever command you installed:</p>-<pre class="language-bash"><code class="language-bash">claude <span class="token parameter variable">--version</span></code></pre>-<p>A version number means you are done with the terminal. “Command not found” means the install did-not finish, or the terminal is still running from before the install: close the window, open a new-one, and try again.</p>-<div class="callout warn"><span class="ic">The one thing people skip</span><p><strong>You must-actually sign in.</strong> Installing the tool is not the same as logging into it. Run the command-once on its own and complete the sign-in before moving on, or Munder Difflin will find the engine-and every agent will still fail the moment it starts.</p></div>-<h2 id="step-3-install-munder-difflin" tabindex="-1">Step 3: Install Munder Difflin <a class="anchor" href="#step-3-install-munder-difflin" aria-hidden="true">#</a></h2>-<p>Go to <a href="https://munderdiffl.in">munderdiffl.in</a> and download the build for your system, or take it-straight from the-<a href="https://github.com/chaitanyagiri/munder-difflin/releases/latest">latest release on GitHub</a>.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-download.png" alt="The Munder Difflin download page showing builds for macOS, Windows and Linux" loading="lazy" decoding="async" /><figcaption>The download page. Pick the file that matches your computer.</figcaption></figure>-<p><strong>macOS.</strong> Download the universal <code>.dmg</code>, which runs on Apple Silicon and Intel. Open it and drag-Munder Difflin into Applications. The app is signed and notarized by Apple, so the first launch only-shows the usual question about opening an app downloaded from the internet. Click Open.</p>-<p><strong>Windows.</strong> Download the setup <code>.exe</code> and run it. If Windows SmartScreen shows a warning, click-More info, then Run anyway. If the installer will not start at all, download the portable <code>.exe</code>-from the release instead, which needs no installation.</p>-<p><strong>Linux.</strong> Download the <code>.AppImage</code>, make it executable (<code>chmod +x</code> on the file) and run it.</p>-<p><strong>Want to be sure the file is genuine?</strong> Every release carries a <code>SHA256SUMS.txt</code>. Run-<code>shasum -a 256</code> on the file you downloaded (or <code>Get-FileHash</code> in Windows PowerShell) and check that-the value matches that file’s line.</p>-<h2 id="step-4-onboarding-six-screens" tabindex="-1">Step 4: Onboarding, six screens <a class="anchor" href="#step-4-onboarding-six-screens" aria-hidden="true">#</a></h2>-<p>Open the app. It walks you through setup. There are two intro screens and then four numbered steps.</p>-<h3 id="screen-1-who-are-you" tabindex="-1">Screen 1: Who are you? <a class="anchor" href="#screen-1-who-are-you" aria-hidden="true">#</a></h3>-<p>It asks whether you are technical or not.</p>-<p><strong>Pick the non-technical option.</strong> This is not a downgrade. It changes the language of the entire-rest of the app, replacing engine ids, flags and model slugs with plain descriptions. You can change-it later in Settings.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-persona.png" alt="The first onboarding screen asking whether you are technical or non-technical" loading="lazy" decoding="async" /><figcaption>The first onboarding screen. Choosing the non-technical path rewrites the rest of the setup in plain language.</figcaption></figure>-<h3 id="screen-2-meet-your-office" tabindex="-1">Screen 2: Meet your office <a class="anchor" href="#screen-2-meet-your-office" aria-hidden="true">#</a></h3>-<p>An explanation of what you are about to set up. Read it and continue.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-welcome.png" alt="The onboarding screen introducing the office floor and its parts" loading="lazy" decoding="async" /><figcaption>The office tour screen.</figcaption></figure>-<h3 id="step-1-of-4-a-home-for-the-app" tabindex="-1">Step 1 of 4: A home for the app <a class="anchor" href="#step-1-of-4-a-home-for-the-app" aria-hidden="true">#</a></h3>-<p>The app needs an empty folder to keep its own files: its settings, and your agents’ memory.</p>-<p>Something like <code>HarnessAgents</code> in your home folder is fine. It will create it if it does not exist.-This is not where your work goes. It is the app’s own filing cabinet.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-home.png" alt="The onboarding step for choosing an empty home folder for the app" loading="lazy" decoding="async" /><figcaption>Choosing a home folder. Make a new empty one.</figcaption></figure>-<h3 id="step-2-of-4-your-clone" tabindex="-1">Step 2 of 4: Your clone <a class="anchor" href="#step-2-of-4-your-clone" aria-hidden="true">#</a></h3>-<p>Name your clone and pick the engine that powers it. This is where the engine you installed in Step 2-shows up. If it is not listed as installed, the sign-in did not finish, so go back and run the-command once more.</p>-<p>Give your clone the most capable model available to you. It is doing the thinking and delegating,-so it benefits from the extra capability more than the workers do.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-engine.png" alt="The onboarding step for naming your clone and picking its AI engine" loading="lazy" decoding="async" /><figcaption>Naming your clone and picking its engine.</figcaption></figure>-<h3 id="step-3-of-4-your-projects" tabindex="-1">Step 3 of 4: Your projects <a class="anchor" href="#step-3-of-4-your-projects" aria-hidden="true">#</a></h3>-<p>Add the folders you want agents working in. A project is just a folder. It can hold code, documents,-notes, or anything else.</p>-<p>You can skip this and add projects later. If you are adding one now, make it a new empty folder.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-repos.png" alt="The onboarding step for adding a project folder" loading="lazy" decoding="async" /><figcaption>Adding a project. A project is simply a folder.</figcaption></figure>-<h3 id="step-4-of-4-permissions" tabindex="-1">Step 4 of 4: Permissions <a class="anchor" href="#step-4-of-4-permissions" aria-hidden="true">#</a></h3>-<p>The important screen. It asks how much agents can do on their own.</p>-<p><strong>On</strong> means agents carry out tasks without stopping to ask. Smoothest, and how you will eventually-want it.</p>-<p><strong>Off</strong> means agents pause and ask you before changing files or running commands.</p>-<p><strong>Start with off.</strong> Watch a few tasks run, see what the agents actually try to do, and turn it on-when you are comfortable. You can change this any time.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-permissions.png" alt="The onboarding step choosing how much agents can do on their own" loading="lazy" decoding="async" /><figcaption>The autonomy choice. Start with agents asking permission.</figcaption></figure>-<h2 id="step-5-your-first-job" tabindex="-1">Step 5: Your first job <a class="anchor" href="#step-5-your-first-job" aria-hidden="true">#</a></h2>-<p>You are looking at the office floor. Your clone is there. Nobody else yet.</p>-<p>Talk to your clone the way you would talk to a capable new hire who has not met you yet. Be specific-about the outcome, and do not bother specifying the steps.</p>-<p>Weak: <em>“organise my files”</em></p>-<p>Better: <em>“Look in my Documents/receipts folder. Sort every PDF into subfolders by year, based on the-date inside the document, not the filename. Tell me about anything you cannot read.”</em></p>-<p>Your clone will decide it needs a worker, hire one, and hand the job over. You will see a new-character appear at a desk.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-floor.png" alt="The Munder Difflin office floor on a fresh setup, with Michael the orchestrator at his desk and the Command Center panel on the right" loading="lazy" decoding="async" /><figcaption>The office floor on a fresh setup. Michael, your clone, is the only one here yet. The Command Center on the right is where work gets dispatched.</figcaption></figure>-<p>Click any agent to open its terminal and watch it work in real time.</p>-<h3 id="things-people-actually-use-this-for" tabindex="-1">Things people actually use this for <a class="anchor" href="#things-people-actually-use-this-for" aria-hidden="true">#</a></h3>-<p>Not all of these are coding.</p>-<ul>-<li><strong>Documents.</strong> Sort a folder of files, rename them consistently, pull the numbers out of fifty-invoices into one spreadsheet.</li>-<li><strong>Research.</strong> Give an agent a question and a place to write, and let it work through it while you-do something else.</li>-<li><strong>Code.</strong> The obvious one. Fix a bug, add a feature, upgrade a dependency, write the tests.</li>-<li><strong>Repeating chores.</strong> Anything you do every Monday, handed to a schedule instead.</li>-<li><strong>Long jobs.</strong> The ones you keep not starting because they take four hours. Start one before bed.</li>-</ul>-<h2 id="step-6-add-agents-yourself" tabindex="-1">Step 6: Add agents yourself <a class="anchor" href="#step-6-add-agents-yourself" aria-hidden="true">#</a></h2>-<p>Your clone hires people on its own, but you can hire directly too.</p>-<p>Click add agent. You pick a name, an engine, a model and a project. Different agents can run on-different engines, so you can put a free engine on the boring jobs and your paid one on the hard-ones.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-addagent.png" alt="The Add Agent panel showing the identity step, a name field, and a grid of selectable Office characters" loading="lazy" decoding="async" /><figcaption>The add agent panel: identity, workspace, engine, briefing. Yes, you pick which Office character they are.</figcaption></figure>-<p>Agents persist. Close an agent and its memory and inbox reattach when you bring it back, with the-same identity.</p>-<h2 id="step-7-work-that-starts-without-you" tabindex="-1">Step 7: Work that starts without you <a class="anchor" href="#step-7-work-that-starts-without-you" aria-hidden="true">#</a></h2>-<p>Open the triggers tab. Everything here starts work when you are not typing.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-triggers.png" alt="The triggers tab showing schedules, context, webhooks and organisation" loading="lazy" decoding="async" /><figcaption>The triggers tab: schedules, context, webhooks and organisation.</figcaption></figure>-<p><strong>Schedules</strong> run a prompt on a repeating clock. You give it a label, choose who it goes to, and-write the prompt, which is sent word for word on every run. This is the feature that turns the app-from a tool you use into something that runs on its own.</p>-<p>Good first schedule: every morning, ask an agent to check something and write you a short summary.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-schedule.png" alt="The new schedule form with label, recipient and prompt fields" loading="lazy" decoding="async" /><figcaption>Creating a schedule. The prompt is sent word for word on every run.</figcaption></figure>-<p><strong>Context</strong> handles what happens as an agent’s memory fills up (see below).</p>-<p><strong>Webhooks</strong> let an outside system post work in.</p>-<p><strong>Organisation</strong> lets a teammate’s Munder Difflin send work to yours.</p>-<h2 id="step-8-context-and-auto-compaction-briefly" tabindex="-1">Step 8: Context and auto-compaction, briefly <a class="anchor" href="#step-8-context-and-auto-compaction-briefly" aria-hidden="true">#</a></h2>-<p>Every AI model has a limited working memory, called its context. A long-running agent will fill it.-When it is full, the agent gets confused or stops.</p>-<p><strong>Auto-compaction</strong> is the fix. When an agent’s context gets close to full, the app has it summarise-what has happened so far and carry the summary forward instead of the raw history. The agent keeps-going without losing the thread.</p>-<p>You control this under Context in the triggers tab: compact automatically, clear entirely, or leave-it alone. The default is sensible and most people never touch it.</p>-<h2 id="step-9-settings-worth-knowing" tabindex="-1">Step 9: Settings worth knowing <a class="anchor" href="#step-9-settings-worth-knowing" aria-hidden="true">#</a></h2>-<p>Seven sections. These are the ones that matter early:</p>+<p>If the message does not settle it, work up this list and stop at the first step that fixes it. This order is our advice from running the app every day; the app itself only offers the first step.</p>+<ol>+<li><strong>Check again.</strong> Press <strong>Check again</strong> or <strong>Try again</strong> on the screen.</li>+<li><strong>Restart the agent.</strong> Close that agent and start it again.</li>+<li><strong>Reopen the app.</strong> Quit Munder Difflin completely and open it again.</li>+<li><strong>Restart the computer.</strong> The last resort, if step 3 did not help.</li>+</ol>+<figure class="mg" data-scene="fixes"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/fixes.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Four stepping stones cross a small stream on white, between two green banks, each stone a little bigger than the last. They are labelled Check again, restart the agent, reopen the app and restart the computer. Every stone carries the same round restart arrow, drawn around a button, a terminal, an app window and a computer. Above the near bank a crooked card reading Install shows a progress bar snapped in two and a yellow warning mark. A pebble hops from the bank onto the first stone, the arrow spins, the broken half of the bar lifts and drops again, and the stone gets a lilac cross. The second stone goes the same way. On the third stone the bar mends and fills, the card straightens, a tick appears and the pebble stays there. The fourth stone is never needed. Past the far bank a speech bubble reads Still stuck? Ask your orchestrator. Tapping a stone makes that stone the one that works."><figcaption>When an install fails, work up four steps and stop at the first that fixes it: Check again, restart the agent, reopen the app, restart the computer. Tap a stone to hop that far.</figcaption></figure>+<p>Still stuck? Ask your orchestrator: <em>“The Codex install failed. Read the log and tell me why.”</em> If the engine that failed is the one your orchestrator runs on, there is nobody to ask yet: run that engine’s lines from the terminal section above, then press <strong>Check again</strong>. The log is <code>hive/log.jsonl</code> inside your workspace folder, and a crashed agent’s raw output is in <code>hive/crashes</code> beside it.</p>+<p>Three things look like failures and are not:</p><ul>-<li><strong>General.</strong> Change your home folder, switch between technical and plain language, pick your-language. The app runs in English, Simplified Chinese and Arabic.</li>-<li><strong>Prerequisites.</strong> Checks whether your engines are actually installed and logged in. <strong>This is the-first place to look when something is not working.</strong></li>-<li><strong>Agents & Models.</strong> Which engines are available and which models each uses. New models show up-without an app update, because the model pickers read the latest list when the app launches.</li>-<li><strong>Autonomy & Budgets.</strong> The permission choice from onboarding, plus spending limits per agent.-Worth setting a cap before leaving anything running overnight.</li>-<li><strong>Connections.</strong> Slack, webhooks, MCP and the REST API.</li>-<li><strong>Voice.</strong> Talk to your agents out loud.</li>-<li><strong>Memory & Knowledge.</strong> What your agents remember between sessions.</li>+<li><strong>An agent reads “looping”.</strong> That is the circuit breaker warning an agent that repeats itself or spends fast. It is the app working.</li>+<li><strong>An agent’s session was cleared.</strong> By default the app clears a session when the agent’s work switches and the session has run for 60 minutes, to save tokens. A file called <code>handoff.md</code> in the agent’s folder carries the thread across.</li>+<li><strong>Everything is slow with many agents.</strong> A computer with 8 GB of memory struggles with a large team. Run fewer at once.</li></ul>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-settings.png" alt="The Munder Difflin settings panel with the Prerequisites section open, listing uv, git and Node.js as ready" loading="lazy" decoding="async" /><figcaption>The settings panel. Prerequisites is where you check that an engine is really installed.</figcaption></figure>-<h2 id="step-10-focus-mode" tabindex="-1">Step 10: Focus mode <a class="anchor" href="#step-10-focus-mode" aria-hidden="true">#</a></h2>-<p>When you want to stop watching the office and work with one agent, press the fullscreen control on-its terminal. The floor disappears and you get a full-width terminal.</p>-<p>Press <code>Esc</code> to come back.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/shot-focus.png" alt="Munder Difflin in focus mode, showing a single agent's terminal at full width with the agent list on the left" loading="lazy" decoding="async" /><figcaption>Focus mode. One agent, full width, no office.</figcaption></figure>-<h2 id="when-something-goes-wrong" tabindex="-1">When something goes wrong <a class="anchor" href="#when-something-goes-wrong" aria-hidden="true">#</a></h2>-<p><strong>An agent dies the moment it starts.</strong> Almost always the engine is installed but not logged in.-Open Settings, then Prerequisites. Or run the engine’s command once in a terminal and finish the-sign-in.</p>-<p><strong>“Engine not installed” during onboarding, but you installed it.</strong> The app looks for the command on-your system path. Close the app completely and reopen it, since it reads your environment at launch.</p>-<p><strong>Windows blocks the installer.</strong> If SmartScreen flags it, choose More info, then Run anyway, or use-the portable build.</p>-<p><strong>macOS asks whether to open an app downloaded from the internet.</strong> That is normal for anything you-download. Click Open.</p>-<p><strong>Nothing is happening.</strong> Check whether message delivery is paused in the Command Center. Paused-means queued work is being held for every agent, and nothing is lost when you switch it back on.</p>-<h2 id="the-short-version" tabindex="-1">The short version <a class="anchor" href="#the-short-version" aria-hidden="true">#</a></h2>+<h2 id="what-is-the-short-version" tabindex="-1">What is the short version? <a class="anchor" href="#what-is-the-short-version" aria-hidden="true">#</a></h2>+<p>Seven steps, about twenty minutes.</p><ol>-<li>Pick an engine. Antigravity if you want free, Codex if you pay for ChatGPT, Claude Code if you-pay for Claude.</li>-<li>Install it and sign in. One terminal command. The only one.</li>-<li>Download Munder Difflin and install it.</li>-<li>Six onboarding screens. Pick the non-technical path, use a new empty folder, start with agents-asking permission.</li>-<li>Give your clone a real job and watch what happens.</li>+<li>Pick one engine: Claude Code, Codex, or a free one.</li>+<li>Download Munder Difflin and open it.</li>+<li>Press <strong>Open browser to sign in</strong> and sign in.</li>+<li>Name your workspace and your orchestrator.</li>+<li>Press <strong>Continue with the free version</strong>.</li>+<li>If the engine is missing, press <strong>Install</strong>, then sign in to it.</li>+<li>Give your orchestrator a real job in a new empty folder, and watch.</li></ol>-<figure class="fig fig-inline"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/note-2.png" alt="Hand-drawn sketch from “How to Install and Use Munder Difflin”" loading="lazy" decoding="async" /></figure>-<p>The app is <a href="https://github.com/chaitanyagiri/munder-difflin">open source on GitHub</a>. If you get-stuck, the <a href="https://munderdiffl.in">Discord</a> is the fastest place to ask.</p>+<link rel="stylesheet" href="/blog/assets/media/how-to-install-and-use-munder-difflin/motion.css"><script defer src="/blog/assets/media/how-to-install-and-use-munder-difflin/motion.js"></script>@@ -589,28 +434,33 @@ stuck, the <a href="https://munderdiffl.in">Discord</a> is the fastest place to<h2 id="faq">FAQ</h2><div class="qa">- <p class="q">Do I need to know how to code to use Munder Difflin?</p>- <p class="a">No. The onboarding asks whether you are technical or not on its very first screen, and the non-technical path replaces every piece of jargon with plain language. You will open a terminal exactly once, to install and log into your AI engine, and you can copy and paste those commands.</p>+ <p class="q">Do I need to know how to code to install Munder Difflin?</p>+ <p class="a">No. You type two or three commands, to install an AI engine and sign in to it, and you can copy and paste them. If you skip that, the app offers to install a missing engine for you the first time an agent starts.</p></div><div class="qa">- <p class="q">Does Munder Difflin cost anything?</p>- <p class="a">The classic app is free and open source, with unlimited local agents. Pro and Teams are optional paid plans. What you may pay for is the AI engine behind it. Antigravity is available at no charge and OpenCode has a free path, so you can run the whole thing without paying anything. If you already pay for ChatGPT or Claude, you can use those subscriptions instead.</p>+ <p class="q">Do I have to install an AI engine before the app?</p>+ <p class="a">No. Setup marks each engine Installed, Installs on first start or Not installed. An engine marked Installs on first start is installed in the agent's own terminal when you press Install, and the app installs Node first if that engine needs it. You still have to sign in to the engine yourself.</p></div><div class="qa">- <p class="q">Can these agents really change files on my computer?</p>- <p class="a">Yes, and that is the point, so it is worth understanding before you start. An agent can read files, write files and run commands in the folders you give it. You choose how much freedom it has during setup, and you can set it to ask permission before every change.</p>+ <p class="q">Do I need an account?</p>+ <p class="a">Yes. The first launch asks you to sign in through your browser, on the free version as well. The browser does not open by itself: press Open browser to sign in and sign in there. The screen usually moves on by itself; if it does not, paste the code the browser shows you into the app.</p></div><div class="qa"><p class="q">Which operating systems does Munder Difflin run on?</p>- <p class="a">macOS, Windows and Linux. macOS ships as one universal .dmg for Apple Silicon and Intel, Windows 10 and 11 as a setup installer or a portable build, and Linux as an AppImage.</p>+ <p class="a">macOS, Windows and Linux. macOS is one universal build for Apple Silicon and Intel, signed and notarized by Apple. Windows 10 and 11, 64 bit, get a setup installer and a portable build. Linux gets a 64 bit AppImage.</p>+ </div>++ <div class="qa">+ <p class="q">Can these agents really change files on my computer?</p>+ <p class="a">Yes, and that is the point, so understand it before you start. An agent starts in the folder you give it and can read files, write files and run commands. It is not locked in there: it can reach anything your account can. Start with a new empty folder and watch what the agents do before you point one at anything you care about.</p></div><div class="qa">- <p class="q">Is my code or data sent anywhere?</p>- <p class="a">The app runs on your machine and stores its files there. Your prompts and the files an agent reads do go to whichever AI engine you picked, the same as if you used that tool directly. Anonymous usage telemetry is opt-out and every event is listed publicly in TELEMETRY.md.</p>+ <p class="q">Is the free version a separate download from Pro?</p>+ <p class="a">No. There is one installer. The free version is the Classic view, and a licence unlocks the Pro view in the same app. On first launch, press Continue with the free version to reach Classic.</p></div></section>@@ -620,7 +470,7 @@ stuck, the <a href="https://munderdiffl.in">Discord</a> is the fastest place to<hr /><div class="share"><span class="lbl">Tags</span>- <a class="tag" href="/blog/tags/guides/">Guides</a><a class="tag" href="/blog/tags/getting-started/">Getting Started</a><a class="tag" href="/blog/tags/tutorial/">Tutorial</a><a class="tag" href="/blog/tags/non-technical/">Non-Technical</a><a class="tag" href="/blog/tags/automation/">Automation</a>+ <a class="tag" href="/blog/tags/guides/">Guides</a><a class="tag" href="/blog/tags/getting-started/">Getting Started</a><a class="tag" href="/blog/tags/tutorial/">Tutorial</a><a class="tag" href="/blog/tags/non-technical/">Non-Technical</a></div></div>@@ -629,7 +479,7 @@ stuck, the <a href="https://munderdiffl.in">Discord</a> is the fastest place to<aside class="toc-rail" aria-label="Table of contents"><div class="tt">On this page</div><ol>- <li class="lvl-2"><a href="#first-what-is-munder-difflin">First, what is Munder Difflin?</a></li><li class="lvl-2"><a href="#what-is-a-coding-agent-and-how-can-it-touch-my-computer">What is a coding agent, and how can it touch my computer?</a></li><li class="lvl-2"><a href="#step-1-which-ai-engine-will-you-use">Step 1: Which AI engine will you use?</a></li><li class="lvl-3"><a href="#you-do-not-want-to-pay-anything">You do not want to pay anything</a></li><li class="lvl-3"><a href="#you-already-pay-for-chatgpt">You already pay for ChatGPT</a></li><li class="lvl-3"><a href="#you-already-pay-for-claude">You already pay for Claude</a></li><li class="lvl-3"><a href="#the-other-engines">The other engines</a></li><li class="lvl-2"><a href="#step-2-install-your-engine-and-log-in">Step 2: Install your engine, and log in</a></li><li class="lvl-3"><a href="#first-do-you-have-nodejs">First: do you have Node.js?</a></li><li class="lvl-3"><a href="#antigravity-free-no-subscription">Antigravity (free, no subscription)</a></li><li class="lvl-3"><a href="#opencode-free-bring-your-own-model">OpenCode (free, bring your own model)</a></li><li class="lvl-3"><a href="#codex-with-a-chatgpt-subscription">Codex (with a ChatGPT subscription)</a></li><li class="lvl-3"><a href="#claude-code-with-a-claude-subscription">Claude Code (with a Claude subscription)</a></li><li class="lvl-3"><a href="#the-other-engines-1">The other engines</a></li><li class="lvl-3"><a href="#check-it-actually-installed">Check it actually installed</a></li><li class="lvl-2"><a href="#step-3-install-munder-difflin">Step 3: Install Munder Difflin</a></li><li class="lvl-2"><a href="#step-4-onboarding-six-screens">Step 4: Onboarding, six screens</a></li><li class="lvl-3"><a href="#screen-1-who-are-you">Screen 1: Who are you?</a></li><li class="lvl-3"><a href="#screen-2-meet-your-office">Screen 2: Meet your office</a></li><li class="lvl-3"><a href="#step-1-of-4-a-home-for-the-app">Step 1 of 4: A home for the app</a></li><li class="lvl-3"><a href="#step-2-of-4-your-clone">Step 2 of 4: Your clone</a></li><li class="lvl-3"><a href="#step-3-of-4-your-projects">Step 3 of 4: Your projects</a></li><li class="lvl-3"><a href="#step-4-of-4-permissions">Step 4 of 4: Permissions</a></li><li class="lvl-2"><a href="#step-5-your-first-job">Step 5: Your first job</a></li><li class="lvl-3"><a href="#things-people-actually-use-this-for">Things people actually use this for</a></li><li class="lvl-2"><a href="#step-6-add-agents-yourself">Step 6: Add agents yourself</a></li><li class="lvl-2"><a href="#step-7-work-that-starts-without-you">Step 7: Work that starts without you</a></li><li class="lvl-2"><a href="#step-8-context-and-auto-compaction-briefly">Step 8: Context and auto-compaction, briefly</a></li><li class="lvl-2"><a href="#step-9-settings-worth-knowing">Step 9: Settings worth knowing</a></li><li class="lvl-2"><a href="#step-10-focus-mode">Step 10: Focus mode</a></li><li class="lvl-2"><a href="#when-something-goes-wrong">When something goes wrong</a></li><li class="lvl-2"><a href="#the-short-version">The short version</a></li>+ <li class="lvl-2"><a href="#what-is-munder-difflin">What is Munder Difflin?</a></li><li class="lvl-2"><a href="#what-is-a-coding-agent-and-what-can-it-touch-on-your-computer">What is a coding agent, and what can it touch on your computer?</a></li><li class="lvl-2"><a href="#what-do-you-need-before-you-install">What do you need before you install?</a></li><li class="lvl-2"><a href="#which-ai-engine-should-you-pick">Which AI engine should you pick?</a></li><li class="lvl-2"><a href="#how-do-you-install-an-engine-and-sign-in">How do you install an engine and sign in?</a></li><li class="lvl-3"><a href="#or-let-the-app-install-it">Or let the app install it</a></li><li class="lvl-2"><a href="#how-do-you-download-and-install-munder-difflin">How do you download and install Munder Difflin?</a></li><li class="lvl-2"><a href="#what-happens-the-first-time-you-open-it">What happens the first time you open it?</a></li><li class="lvl-2"><a href="#what-do-you-do-in-your-first-ten-minutes">What do you do in your first ten minutes?</a></li><li class="lvl-2"><a href="#what-is-the-ide-button-for">What is the IDE button for?</a></li><li class="lvl-2"><a href="#what-can-you-add-after-setup">What can you add after setup?</a></li><li class="lvl-2"><a href="#what-do-you-do-when-an-install-fails">What do you do when an install fails?</a></li><li class="lvl-2"><a href="#what-is-the-short-version">What is the short version?</a></li></ol></aside>@@ -643,50 +493,50 @@ stuck, the <a href="https://munderdiffl.in">Discord</a> is the fastest place to<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-install-codex-cli/index.html b/docs/blog/how-to-install-codex-cli/index.htmlindex d2b4e94e..0f892472 100644--- a/docs/blog/how-to-install-codex-cli/index.html+++ b/docs/blog/how-to-install-codex-cli/index.html@@ -347,50 +347,50 @@ $ ls ~/.codex/packages/standalone/releases<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-install-gemini-cli/index.html b/docs/blog/how-to-install-gemini-cli/index.htmlindex 4f95ffdc..ad01a7f4 100644--- a/docs/blog/how-to-install-gemini-cli/index.html+++ b/docs/blog/how-to-install-gemini-cli/index.html@@ -304,50 +304,50 @@ gemini</code></pre><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-run-multiple-claude-code-agents/index.html b/docs/blog/how-to-run-multiple-claude-code-agents/index.htmlindex f13a5532..b2a6c356 100644--- a/docs/blog/how-to-run-multiple-claude-code-agents/index.html+++ b/docs/blog/how-to-run-multiple-claude-code-agents/index.html@@ -270,50 +270,50 @@ free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-talk-to-your-orchestrator/index.html b/docs/blog/how-to-talk-to-your-orchestrator/index.htmlindex 340f2144..cf551484 100644--- a/docs/blog/how-to-talk-to-your-orchestrator/index.html+++ b/docs/blog/how-to-talk-to-your-orchestrator/index.html@@ -293,50 +293,50 @@ Three goals in a trench coat. Split them — the task board handles dependencies<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-use-claude-code-plan-mode/index.html b/docs/blog/how-to-use-claude-code-plan-mode/index.htmlindex 776da618..bf14937e 100644--- a/docs/blog/how-to-use-claude-code-plan-mode/index.html+++ b/docs/blog/how-to-use-claude-code-plan-mode/index.html@@ -413,50 +413,50 @@ The user wants a `multiply` function in `hello.py`, next to the existing `add` a<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-use-claude-code/index.html b/docs/blog/how-to-use-claude-code/index.htmlindex 1f4ee071..5e4ebd61 100644--- a/docs/blog/how-to-use-claude-code/index.html+++ b/docs/blog/how-to-use-claude-code/index.html@@ -374,50 +374,50 @@ claude</code></pre><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/how-to-use-gemini-4-argon/index.html b/docs/blog/how-to-use-gemini-4-argon/index.htmlindex c97622f4..10c12ba5 100644--- a/docs/blog/how-to-use-gemini-4-argon/index.html+++ b/docs/blog/how-to-use-gemini-4-argon/index.html@@ -316,50 +316,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/codex-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/codex-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-05">Oct 5, 2026</time><span class="dot" aria-hidden="true"></span><span>5 min</span></div>- <h3><a href="/blog/codex-pricing/">Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)</a></h3>- <p class="dek">Codex pricing checked 5 Oct 2026: which ChatGPT plans include Codex, US prices from $0 to $500, how usage limits are counted, credits and API key rates.</p>+ <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>+ <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/codex-pricing/" aria-label="Read: Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)">Read →</a>+ <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a></div></div></article>diff --git a/docs/blog/how-to-use-the-built-in-monaco-ide/index.html b/docs/blog/how-to-use-the-built-in-monaco-ide/index.htmlindex 1e1cb5fd..28d92957 100644--- a/docs/blog/how-to-use-the-built-in-monaco-ide/index.html+++ b/docs/blog/how-to-use-the-built-in-monaco-ide/index.html@@ -291,50 +291,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/human-in-the-loop-approving-ai-agents/index.html b/docs/blog/human-in-the-loop-approving-ai-agents/index.htmlindex 1d6509ae..1dfeea0a 100644--- a/docs/blog/human-in-the-loop-approving-ai-agents/index.html+++ b/docs/blog/human-in-the-loop-approving-ai-agents/index.html@@ -326,50 +326,50 @@ you talk to? You can <a href="/#install">download Munder Difflin</a> free; it’<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/index.html b/docs/blog/index.htmlindex ff59ee72..9cdca5a4 100644--- a/docs/blog/index.html+++ b/docs/blog/index.html@@ -135,11 +135,11 @@building a self-coordinating hive of agents with a GOD orchestrator you talk to.</p><div class="filters" role="navigation" aria-label="Topics">- <a class="chip" aria-current="page">All<span class="ct">212</span></a>+ <a class="chip" aria-current="page">All<span class="ct">213</span></a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>@@ -183,11 +183,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>@@ -197,13 +197,29 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article>diff --git a/docs/blog/launching-munder-difflin-v0-3-2/index.html b/docs/blog/launching-munder-difflin-v0-3-2/index.htmlindex da188205..f8642f13 100644--- a/docs/blog/launching-munder-difflin-v0-3-2/index.html+++ b/docs/blog/launching-munder-difflin-v0-3-2/index.html@@ -319,7 +319,7 @@<footer class="post-foot wrap"><div class="prevnext"><a class="pn prev" href="/blog/run-munder-difflin-on-a-mac-mini/"><span class="k">← Older</span><span class="t">Run Munder Difflin Locally on a Mac Mini</span></a>- <a class="pn next" href="/blog/your-first-hour-with-munder-difflin/"><span class="k">Newer →</span><span class="t">Your First Hour With Munder Difflin: Onboarding to a Pro Office</span></a>+ <a class="pn next" href="/blog/your-first-hour-with-munder-difflin/"><span class="k">Newer →</span><span class="t">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</span></a></div><section class="related"><h2>Related in Story</h2>diff --git a/docs/blog/manage-multiple-claude-code-sessions/index.html b/docs/blog/manage-multiple-claude-code-sessions/index.htmlindex 9ae2f79c..68c45b72 100644--- a/docs/blog/manage-multiple-claude-code-sessions/index.html+++ b/docs/blog/manage-multiple-claude-code-sessions/index.html@@ -324,50 +324,50 @@ isolated agents, shared memory, and a single committer, all on your own machine.<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/mcp-and-skills-in-a-hive/index.html b/docs/blog/mcp-and-skills-in-a-hive/index.htmlindex 81d3d76b..d271e94a 100644--- a/docs/blog/mcp-and-skills-in-a-hive/index.html+++ b/docs/blog/mcp-and-skills-in-a-hive/index.html@@ -339,50 +339,50 @@ existing setup into a coordinated team of agents; it’s free and open source.</<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/mcp-security-tool-poisoning/index.html b/docs/blog/mcp-security-tool-poisoning/index.htmlindex 8571cb72..3385edb2 100644--- a/docs/blog/mcp-security-tool-poisoning/index.html+++ b/docs/blog/mcp-security-tool-poisoning/index.html@@ -322,56 +322,56 @@ execution. Do that and MCP is what it should be: a powerful, <em>bounded</em> wa<footer class="post-foot wrap"><div class="prevnext"><a class="pn prev" href="/blog/multi-agent-orchestration-patterns/"><span class="k">← Older</span><span class="t">Multi-Agent Orchestration Patterns: Which Topology, and When</span></a>- <a class="pn next" href="/blog/how-to-install-and-use-munder-difflin/"><span class="k">Newer →</span><span class="t">How to Install and Use Munder Difflin: A Beginner's Guide</span></a>+ <a class="pn next" href="/blog/how-to-install-and-use-munder-difflin/"><span class="k">Newer →</span><span class="t">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</span></a></div><section class="related"><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/munder-difflin-v0-2-4-feature-walkthrough/index.html b/docs/blog/munder-difflin-v0-2-4-feature-walkthrough/index.htmlindex 97510cc1..120fcd40 100644--- a/docs/blog/munder-difflin-v0-2-4-feature-walkthrough/index.html+++ b/docs/blog/munder-difflin-v0-2-4-feature-walkthrough/index.html@@ -304,50 +304,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/prompt-caching-for-ai-agents/index.html b/docs/blog/prompt-caching-for-ai-agents/index.htmlindex f3a33c04..8c6e973d 100644--- a/docs/blog/prompt-caching-for-ai-agents/index.html+++ b/docs/blog/prompt-caching-for-ai-agents/index.html@@ -358,50 +358,50 @@ that doesn’t waste your tokens; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/run-a-mixed-engine-office/index.html b/docs/blog/run-a-mixed-engine-office/index.htmlindex 0e40c9df..f0a485f0 100644--- a/docs/blog/run-a-mixed-engine-office/index.html+++ b/docs/blog/run-a-mixed-engine-office/index.html@@ -286,50 +286,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/run-ai-agent-hive-from-slack-setup/index.html b/docs/blog/run-ai-agent-hive-from-slack-setup/index.htmlindex 20ef7f92..f84e2786 100644--- a/docs/blog/run-ai-agent-hive-from-slack-setup/index.html+++ b/docs/blog/run-ai-agent-hive-from-slack-setup/index.html@@ -385,50 +385,50 @@ it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/run-munder-difflin-on-a-mac-mini/index.html b/docs/blog/run-munder-difflin-on-a-mac-mini/index.htmlindex 61adee36..7df78bed 100644--- a/docs/blog/run-munder-difflin-on-a-mac-mini/index.html+++ b/docs/blog/run-munder-difflin-on-a-mac-mini/index.html@@ -419,50 +419,50 @@ machine awake while at least one agent is running.</li><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/run-munder-difflin-on-open-models/index.html b/docs/blog/run-munder-difflin-on-open-models/index.htmlindex beab3932..9873b0ee 100644--- a/docs/blog/run-munder-difflin-on-open-models/index.html+++ b/docs/blog/run-munder-difflin-on-open-models/index.html@@ -535,50 +535,50 @@ and want the sizing walkthrough? Read the <a href="/blog/run-munder-difflin-on-a<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/security-for-ai-coding-agents/index.html b/docs/blog/security-for-ai-coding-agents/index.htmlindex e1f034bc..8051dd3d 100644--- a/docs/blog/security-for-ai-coding-agents/index.html+++ b/docs/blog/security-for-ai-coding-agents/index.html@@ -323,50 +323,50 @@ and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/small-team-ai-agents-playbook/index.html b/docs/blog/small-team-ai-agents-playbook/index.htmlindex 51633a4d..39c3f85d 100644--- a/docs/blog/small-team-ai-agents-playbook/index.html+++ b/docs/blog/small-team-ai-agents-playbook/index.html@@ -328,50 +328,50 @@ it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/tags/ai-agents/index.html b/docs/blog/tags/ai-agents/index.htmlindex 3d583564..6b5c6847 100644--- a/docs/blog/tags/ai-agents/index.html+++ b/docs/blog/tags/ai-agents/index.html@@ -105,11 +105,27 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#AI Agents</h1>- <p class="lead">27 posts tagged <strong>AI Agents</strong>.</p>+ <p class="lead">28 posts tagged <strong>AI Agents</strong>.</p></header><section class="wrap"><div class="post-grid">- <article class="card t-concepts">+ <article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a>+ </div>+ </div>+</article>+<article class="card t-concepts"><a class="thumb t-concepts" href="/blog/gemini-4-argon-inside-google/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-inside-google/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">diff --git a/docs/blog/tags/automation/index.html b/docs/blog/tags/automation/index.htmlindex 55b8d18e..41176860 100644--- a/docs/blog/tags/automation/index.html+++ b/docs/blog/tags/automation/index.html@@ -105,7 +105,7 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Automation</h1>- <p class="lead">16 posts tagged <strong>Automation</strong>.</p>+ <p class="lead">15 posts tagged <strong>Automation</strong>.</p></header><section class="wrap"><div class="post-grid">@@ -269,22 +269,6 @@</div></div></article>-<article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-install-and-use-munder-difflin/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-install-and-use-munder-difflin/hero.png" alt="" loading="lazy" decoding="async" /></a>- <div class="body">- <div class="meta">- <time datetime="2026-06-05">Jun 5, 2026</time>- <span class="dot" aria-hidden="true"></span>- <span>14 min</span><span class="pin">Pinned</span>- </div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>- <div class="foot">- <span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>- </div>- </div>-</article><article class="card t-guides"><a class="thumb t-guides" href="/blog/claude-code-automation-while-you-sleep/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/claude-code-automation-while-you-sleep/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body">diff --git a/docs/blog/tags/getting-started/index.html b/docs/blog/tags/getting-started/index.htmlindex 3cbf8f73..182d8ccc 100644--- a/docs/blog/tags/getting-started/index.html+++ b/docs/blog/tags/getting-started/index.html@@ -105,11 +105,27 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Getting Started</h1>- <p class="lead">19 posts tagged <strong>Getting Started</strong>.</p>+ <p class="lead">20 posts tagged <strong>Getting Started</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides"><a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">@@ -291,13 +307,13 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a></div></div></article>@@ -309,11 +325,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/tags/guides/index.html b/docs/blog/tags/guides/index.htmlindex 8256eb85..fd26f1ba 100644--- a/docs/blog/tags/guides/index.html+++ b/docs/blog/tags/guides/index.html@@ -105,11 +105,27 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Guides</h1>- <p class="lead">38 posts tagged <strong>Guides</strong>.</p>+ <p class="lead">39 posts tagged <strong>Guides</strong>.</p></header><section class="wrap"><div class="post-grid"><article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides"><a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">@@ -451,13 +467,13 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a></div></div></article>@@ -549,11 +565,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/tags/local-first/index.html b/docs/blog/tags/local-first/index.htmlindex 4faf0d31..2da69cce 100644--- a/docs/blog/tags/local-first/index.html+++ b/docs/blog/tags/local-first/index.html@@ -105,7 +105,7 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Local-First</h1>- <p class="lead">23 posts tagged <strong>Local-First</strong>.</p>+ <p class="lead">22 posts tagged <strong>Local-First</strong>.</p></header><section class="wrap"><div class="post-grid">@@ -317,22 +317,6 @@</div></div></article>-<article class="card t-guides">- <a class="thumb t-guides" href="/blog/your-first-hour-with-munder-difflin/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/hero.png" alt="" loading="lazy" decoding="async" /></a>- <div class="body">- <div class="meta">- <time datetime="2026-07-03">Jul 3, 2026</time>- <span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>- </div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>- <div class="foot">- <span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>- </div>- </div>-</article><article class="card t-guides"><a class="thumb t-guides" href="/blog/run-munder-difflin-on-a-mac-mini/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/run-munder-difflin-on-a-mac-mini/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body">diff --git a/docs/blog/tags/multi-agent/index.html b/docs/blog/tags/multi-agent/index.htmlindex e277754e..eb5d6528 100644--- a/docs/blog/tags/multi-agent/index.html+++ b/docs/blog/tags/multi-agent/index.html@@ -105,11 +105,27 @@<header class="page-head wrap"><p class="eyebrow">Tag</p><h1>#Multi-Agent</h1>- <p class="lead">88 posts tagged <strong>Multi-Agent</strong>.</p>+ <p class="lead">89 posts tagged <strong>Multi-Agent</strong>.</p></header><section class="wrap"><div class="post-grid">- <article class="card t-comparisons">+ <article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a>+ </div>+ </div>+</article>+<article class="card t-comparisons"><a class="thumb t-comparisons" href="/blog/munder-difflin-vs-chatgpt-dots/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/munder-difflin-vs-chatgpt-dots/lead-still.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">@@ -467,13 +483,13 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/tags/non-technical/index.html b/docs/blog/tags/non-technical/index.htmlindex 69d4cb95..be6ea518 100644--- a/docs/blog/tags/non-technical/index.html+++ b/docs/blog/tags/non-technical/index.html@@ -117,11 +117,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/tags/onboarding/index.html b/docs/blog/tags/onboarding/index.htmlindex baa4a6b1..5000dbfc 100644--- a/docs/blog/tags/onboarding/index.html+++ b/docs/blog/tags/onboarding/index.html@@ -115,13 +115,13 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/tags/tutorial/index.html b/docs/blog/tags/tutorial/index.htmlindex a6a00908..00731518 100644--- a/docs/blog/tags/tutorial/index.html+++ b/docs/blog/tags/tutorial/index.html@@ -133,11 +133,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>diff --git a/docs/blog/the-lethal-trifecta-for-coding-agents/index.html b/docs/blog/the-lethal-trifecta-for-coding-agents/index.htmlindex 0c5fd068..dd3678d6 100644--- a/docs/blog/the-lethal-trifecta-for-coding-agents/index.html+++ b/docs/blog/the-lethal-trifecta-for-coding-agents/index.html@@ -346,50 +346,50 @@ secrets; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/the-multi-agent-cost-playbook/index.html b/docs/blog/the-multi-agent-cost-playbook/index.htmlindex 576eddcb..b7c1be53 100644--- a/docs/blog/the-multi-agent-cost-playbook/index.html+++ b/docs/blog/the-multi-agent-cost-playbook/index.html@@ -328,50 +328,50 @@ to run a fleet that doesn’t bankrupt you — it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/topics/comparisons/index.html b/docs/blog/topics/comparisons/index.htmlindex b9e68e2e..81e90796 100644--- a/docs/blog/topics/comparisons/index.html+++ b/docs/blog/topics/comparisons/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/" aria-current="page">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/concepts/index.html b/docs/blog/topics/concepts/index.htmlindex 33446b93..4bc4d267 100644--- a/docs/blog/topics/concepts/index.html+++ b/docs/blog/topics/concepts/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/" aria-current="page">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/" aria-current="page">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/guides/index.html b/docs/blog/topics/guides/index.htmlindex 5fb5fe7f..fa57197e 100644--- a/docs/blog/topics/guides/index.html+++ b/docs/blog/topics/guides/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/" aria-current="page">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/" aria-current="page">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>@@ -145,11 +145,11 @@<span class="dot" aria-hidden="true"></span><span>14 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install and Use Munder Difflin: A Beginner's Guide</a></h3>- <p class="dek">A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.</p>+ <h3><a href="/blog/how-to-install-and-use-munder-difflin/">How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office</a></h3>+ <p class="dek">Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install and Use Munder Difflin: A Beginner's Guide">Read →</a>+ <a class="more" href="/blog/how-to-install-and-use-munder-difflin/" aria-label="Read: How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office">Read →</a></div></div></article>@@ -159,13 +159,29 @@<div class="meta"><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true"></span>- <span>13 min</span><span class="pin">Pinned</span>+ <span>10 min</span><span class="pin">Pinned</span>+ </div>+ <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</a></h3>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p>+ <div class="foot">+ <span class="tag kind">Guides</span>+ <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office">Read →</a>+ </div>+ </div>+</article>+<article class="card t-guides">+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <div class="body">+ <div class="meta">+ <time datetime="2026-10-14">Oct 14, 2026</time>+ <span class="dot" aria-hidden="true"></span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/your-first-hour-with-munder-difflin/">Your First Hour With Munder Difflin: Onboarding to a Pro Office</a></h3>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/your-first-hour-with-munder-difflin/" aria-label="Read: Your First Hour With Munder Difflin: Onboarding to a Pro Office">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article>diff --git a/docs/blog/topics/index.html b/docs/blog/topics/index.htmlindex a9336f43..65d780aa 100644--- a/docs/blog/topics/index.html+++ b/docs/blog/topics/index.html@@ -113,19 +113,19 @@<section class="wrap"><div class="post-grid">- <a class="card" href="/blog/topics/concepts/" style="text-decoration:none">- <span class="thumb t-concepts" aria-hidden="true"><img src="/blog/assets/media/topics/concepts.png" alt="" width="960" height="540" loading="lazy" decoding="async" /></span>+ <a class="card" href="/blog/topics/guides/" style="text-decoration:none">+ <span class="thumb t-guides" aria-hidden="true"><img src="/blog/assets/media/topics/guides.png" alt="" width="960" height="540" loading="lazy" decoding="async" /></span><div class="body">- <h3 style="color:var(--ink)">concepts</h3>- <p class="dek">49 posts.</p>+ <h3 style="color:var(--ink)">guides</h3>+ <p class="dek">50 posts.</p><div class="foot"><span class="more">View topic →</span></div></div></a>- <a class="card" href="/blog/topics/guides/" style="text-decoration:none">- <span class="thumb t-guides" aria-hidden="true"><img src="/blog/assets/media/topics/guides.png" alt="" width="960" height="540" loading="lazy" decoding="async" /></span>+ <a class="card" href="/blog/topics/concepts/" style="text-decoration:none">+ <span class="thumb t-concepts" aria-hidden="true"><img src="/blog/assets/media/topics/concepts.png" alt="" width="960" height="540" loading="lazy" decoding="async" /></span><div class="body">- <h3 style="color:var(--ink)">guides</h3>+ <h3 style="color:var(--ink)">concepts</h3><p class="dek">49 posts.</p><div class="foot"><span class="more">View topic →</span></div></div>@@ -200,7 +200,7 @@<div class="filters" style="padding-top:36px"><span class="eyebrow" style="align-self:center">Tags</span>- <a class="chip" href="/blog/tags/multi-agent/">Multi-Agent<span class="ct">88</span></a>+ <a class="chip" href="/blog/tags/multi-agent/">Multi-Agent<span class="ct">89</span></a><a class="chip" href="/blog/tags/claude-code/">Claude Code<span class="ct">86</span></a>@@ -210,19 +210,19 @@<a class="chip" href="/blog/tags/comparisons/">Comparisons<span class="ct">42</span></a>- <a class="chip" href="/blog/tags/guides/">Guides<span class="ct">38</span></a>+ <a class="chip" href="/blog/tags/guides/">Guides<span class="ct">39</span></a>- <a class="chip" href="/blog/tags/ai-agents/">AI Agents<span class="ct">27</span></a>+ <a class="chip" href="/blog/tags/ai-agents/">AI Agents<span class="ct">28</span></a><a class="chip" href="/blog/tags/internals/">Internals<span class="ct">26</span></a>- <a class="chip" href="/blog/tags/local-first/">Local-First<span class="ct">23</span></a>-<a class="chip" href="/blog/tags/cost/">Cost<span class="ct">23</span></a>- <a class="chip" href="/blog/tags/story/">Story<span class="ct">20</span></a>+ <a class="chip" href="/blog/tags/local-first/">Local-First<span class="ct">22</span></a>- <a class="chip" href="/blog/tags/getting-started/">Getting Started<span class="ct">19</span></a>+ <a class="chip" href="/blog/tags/getting-started/">Getting Started<span class="ct">20</span></a>++ <a class="chip" href="/blog/tags/story/">Story<span class="ct">20</span></a><a class="chip" href="/blog/tags/orchestration/">Orchestration<span class="ct">19</span></a>@@ -232,10 +232,10 @@<a class="chip" href="/blog/tags/engines/">Engines<span class="ct">17</span></a>- <a class="chip" href="/blog/tags/automation/">Automation<span class="ct">16</span></a>-<a class="chip" href="/blog/tags/release/">Release<span class="ct">16</span></a>+ <a class="chip" href="/blog/tags/automation/">Automation<span class="ct">15</span></a>+<a class="chip" href="/blog/tags/skills/">Skills<span class="ct">13</span></a><a class="chip" href="/blog/tags/tools/">Tools<span class="ct">12</span></a>diff --git a/docs/blog/topics/internals/index.html b/docs/blog/topics/internals/index.htmlindex 31db8f12..106b1e6b 100644--- a/docs/blog/topics/internals/index.html+++ b/docs/blog/topics/internals/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/memory/index.html b/docs/blog/topics/memory/index.htmlindex cb4b4108..f80a8e27 100644--- a/docs/blog/topics/memory/index.html+++ b/docs/blog/topics/memory/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/news/index.html b/docs/blog/topics/news/index.htmlindex bee88260..9acfba64 100644--- a/docs/blog/topics/news/index.html+++ b/docs/blog/topics/news/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/orchestration/index.html b/docs/blog/topics/orchestration/index.htmlindex 2c24d412..60371c20 100644--- a/docs/blog/topics/orchestration/index.html+++ b/docs/blog/topics/orchestration/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/story/index.html b/docs/blog/topics/story/index.htmlindex 925214d5..669616ae 100644--- a/docs/blog/topics/story/index.html+++ b/docs/blog/topics/story/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/topics/use-cases/index.html b/docs/blog/topics/use-cases/index.htmlindex d5738f57..630b7bae 100644--- a/docs/blog/topics/use-cases/index.html+++ b/docs/blog/topics/use-cases/index.html@@ -114,9 +114,9 @@<div class="filters" role="navigation" aria-label="Topics"><a class="chip" href="/blog/">All</a>- <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/guides/">guides<span class="ct">50</span></a>- <a class="chip" href="/blog/topics/guides/">guides<span class="ct">49</span></a>+ <a class="chip" href="/blog/topics/concepts/">concepts<span class="ct">49</span></a><a class="chip" href="/blog/topics/comparisons/">comparisons<span class="ct">42</span></a>diff --git a/docs/blog/trigger-ai-agents-from-github-webhook/index.html b/docs/blog/trigger-ai-agents-from-github-webhook/index.htmlindex 424ff163..0cf3d70b 100644--- a/docs/blog/trigger-ai-agents-from-github-webhook/index.html+++ b/docs/blog/trigger-ai-agents-from-github-webhook/index.html@@ -376,50 +376,50 @@ open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/trigger-ai-agents-from-slack/index.html b/docs/blog/trigger-ai-agents-from-slack/index.htmlindex 9568c236..d9e14992 100644--- a/docs/blog/trigger-ai-agents-from-slack/index.html+++ b/docs/blog/trigger-ai-agents-from-slack/index.html@@ -324,50 +324,50 @@ it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/what-is-harness-engineering/index.html b/docs/blog/what-is-harness-engineering/index.htmlindex 1fc0c1f3..5bfafb15 100644--- a/docs/blog/what-is-harness-engineering/index.html+++ b/docs/blog/what-is-harness-engineering/index.html@@ -281,7 +281,7 @@</div><footer class="post-foot wrap"><div class="prevnext">- <a class="pn prev" href="/blog/your-first-hour-with-munder-difflin/"><span class="k">← Older</span><span class="t">Your First Hour With Munder Difflin: Onboarding to a Pro Office</span></a>+ <a class="pn prev" href="/blog/your-first-hour-with-munder-difflin/"><span class="k">← Older</span><span class="t">Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</span></a><a class="pn next" href="/blog/what-is-an-agent-development-environment/"><span class="k">Newer →</span><span class="t">What Is an Agent Development Environment (ADE)?</span></a></div><section class="related">diff --git a/docs/blog/whats-new-in-agentic-ai-june-2026/index.html b/docs/blog/whats-new-in-agentic-ai-june-2026/index.htmlindex 9a0ebb97..025aa634 100644--- a/docs/blog/whats-new-in-agentic-ai-june-2026/index.html+++ b/docs/blog/whats-new-in-agentic-ai-june-2026/index.html@@ -383,50 +383,50 @@ hive run; it’s free and open source.</p><h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/why-does-claude-code-keep-asking-for-permission/index.html b/docs/blog/why-does-claude-code-keep-asking-for-permission/index.htmlindex 37bff2b2..41155bc4 100644--- a/docs/blog/why-does-claude-code-keep-asking-for-permission/index.html+++ b/docs/blog/why-does-claude-code-keep-asking-for-permission/index.html@@ -289,50 +289,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/blog/your-first-hour-with-munder-difflin/index.html b/docs/blog/your-first-hour-with-munder-difflin/index.htmlindex fb245d17..a7f24dbe 100644--- a/docs/blog/your-first-hour-with-munder-difflin/index.html+++ b/docs/blog/your-first-hour-with-munder-difflin/index.html@@ -18,8 +18,8 @@<meta charset="UTF-8" /><meta name="viewport" content="width=device-width, initial-scale=1.0" />-<title>Your First Hour With Munder Difflin: Onboarding to a Pro Office — Munder Difflin Blog</title>-<meta name="description" content="Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email." />+<title>Munder Difflin Onboarding: Your First Hour With 0.5.6</title>+<meta name="description" content="Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens." /><link rel="canonical" href="https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/" /><link rel="icon" type="image/png" href="https://munderdiffl.in/logo.png" />@@ -29,13 +29,13 @@<!-- Open Graph --><meta property="og:site_name" content="Munder Difflin Blog" /><meta property="og:type" content="article" />-<meta property="og:title" content="Your First Hour With Munder Difflin: Onboarding to a Pro Office — Munder Difflin Blog" />-<meta property="og:description" content="Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email." />+<meta property="og:title" content="Munder Difflin Onboarding: Your First Hour With 0.5.6" />+<meta property="og:description" content="Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens." /><meta property="og:url" content="https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/" /><meta property="og:image" content="https://munderdiffl.in/blog/assets/media/your-first-hour-with-munder-difflin/hero.png" /><meta property="og:image:width" content="1600" /><meta property="og:image:height" content="900" />-<meta property="og:image:alt" content="Michael, the orchestrator, in the Pro app after dispatching four tasks to his agents" />+<meta property="og:image:alt" content="Title card on white. Large ink type reads Your First Hour With Munder Difflin 0.5.6, with the line From first sign in to a working office and a pill reading Onboarding guide, beside a blue stopwatch whose yellow ring has gone once all the way round as it rings." /><meta property="og:locale" content="en_US" /><meta property="article:published_time" content="2026-07-03T00:00:00.000Z" /><meta property="article:author" content="Chaitanya Giri" />@@ -43,15 +43,14 @@<meta property="article:tag" content="Onboarding" /><meta property="article:tag" content="Getting Started" /><meta property="article:tag" content="Multi-Agent" />-<meta property="article:tag" content="Local-First" /><!-- Twitter --><meta name="twitter:card" content="summary_large_image" />-<meta name="twitter:title" content="Your First Hour With Munder Difflin: Onboarding to a Pro Office — Munder Difflin Blog" />-<meta name="twitter:description" content="Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email." />+<meta name="twitter:title" content="Munder Difflin Onboarding: Your First Hour With 0.5.6" />+<meta name="twitter:description" content="Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens." /><meta name="twitter:image" content="https://munderdiffl.in/blog/assets/media/your-first-hour-with-munder-difflin/hero.png" />-<meta name="twitter:image:alt" content="Michael, the orchestrator, in the Pro app after dispatching four tasks to his agents" />+<meta name="twitter:image:alt" content="Title card on white. Large ink type reads Your First Hour With Munder Difflin 0.5.6, with the line From first sign in to a working office and a pill reading Onboarding guide, beside a blue stopwatch whose yellow ring has gone once all the way round as it rings." /><link rel="alternate" type="application/atom+xml" title="Munder Difflin Blog" href="https://munderdiffl.in/blog/feed.xml" /><link rel="preconnect" href="https://fonts.googleapis.com" />@@ -63,11 +62,11 @@{"@context": "https://schema.org","@type": "BlogPosting",- "headline": "Your First Hour With Munder Difflin: Onboarding to a Pro Office",- "description": "Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.",+ "headline": "Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office",+ "description": "Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.","image": ["https://munderdiffl.in/blog/assets/media/your-first-hour-with-munder-difflin/hero.png"],"datePublished": "2026-07-03T00:00:00.000Z",- "dateModified": "2026-09-14T00:00:00.000Z",+ "dateModified": "2026-10-14T00:00:00.000Z","author": { "@type": "Person", "name": "Chaitanya Giri" },"publisher": {"@type": "Organization",@@ -75,7 +74,7 @@"logo": { "@type": "ImageObject", "url": "https://munderdiffl.in/logo.png" }},"mainEntityOfPage": { "@type": "WebPage", "@id": "https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/" },- "keywords": "munder difflin onboarding, getting started with munder difflin, munder difflin tutorial, munder difflin pro, munder difflin setup, scheduled agent automations, munder difflin stapler",+ "keywords": "munder difflin onboarding, getting started with munder difflin, munder difflin tutorial, munder difflin pro, munder difflin temps, munder difflin automations, munder difflin stapler, munder difflin first hour","articleSection": "Guides","inLanguage": "en-US","url": "https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/"@@ -88,7 +87,7 @@"itemListElement": [{ "@type": "ListItem", "position": 1, "name": "Home", "item": "https://munderdiffl.in/" },{ "@type": "ListItem", "position": 2, "name": "Blog", "item": "https://munderdiffl.in/blog/" },- { "@type": "ListItem", "position": 3, "name": "Your First Hour With Munder Difflin: Onboarding to a Pro Office", "item": "https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/" }+ { "@type": "ListItem", "position": 3, "name": "Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office", "item": "https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/" }]}</script>@@ -98,12 +97,12 @@"@context": "https://schema.org","@type": "FAQPage","mainEntity": [- { "@type": "Question", "name": "How long does it take to get Munder Difflin running?", "acceptedAnswer": { "@type": "Answer", "text": "About ten minutes if Claude Code, Codex or OpenCode is already installed and signed in: download, sign in, answer four setup steps. Allow the rest of the hour to hire a small team and set up your first automations." } },- { "@type": "Question", "name": "Do I need an account to use Munder Difflin?", "acceptedAnswer": { "@type": "Answer", "text": "Yes, in 0.5.2. The first launch asks you to sign in with an email address in your browser, on the free plan as well. Pro then needs a licence key on top of that account." } },- { "@type": "Question", "name": "What does Pro add over the free version?", "acceptedAnswer": { "@type": "Answer", "text": "The free version is the classic office. Pro adds a single window workspace with screens for the orchestrator, agents, agent rooms, tasks, inbox, automations, memory, capabilities and temps, plus the Stapler. It runs on one machine at a time; the pricing page has the rest." } },- { "@type": "Question", "name": "Can one Pro office mix Claude Code, Codex and OpenCode agents?", "acceptedAnswer": { "@type": "Answer", "text": "Yes. Each agent picks its own provider and model when you add it, and can have its own token cap. Claude Code and Codex use their own logins; OpenCode can use a provider key you save under Settings." } },- { "@type": "Question", "name": "Does the office keep working when I close my laptop?", "acceptedAnswer": { "@type": "Answer", "text": "No. Agents, automations, Slack and webhooks all run on your machine, so they stop when the app closes or the laptop sleeps. A schedule missed while the laptop slept runs once when the app wakes; a weekly one only if that is within six hours of its slot." } },- { "@type": "Question", "name": "Is the Stapler free?", "acceptedAnswer": { "@type": "Answer", "text": "No. The Stapler is Pro only, and it is off until you turn on **Show the stapler** on the Stapler screen." } }+ { "@type": "Question", "name": "How long does it take to get Munder Difflin running?", "acceptedAnswer": { "@type": "Answer", "text": "Setup is four short steps when an AI engine is already installed and signed in: download, sign in, name your workspace and your orchestrator. The rest of the hour goes on a first job, a small team and the first work that starts without you." } },+ { "@type": "Question", "name": "Do I need Pro for this first hour?", "acceptedAnswer": { "@type": "Answer", "text": "No. Agents, schedules, the context rules, webhooks, Slack and temps are in both views. The Pro view adds one window with screens for Tasks, Inbox, Automations, Memory and Capabilities, plus the Stapler, a small window above every app." } },+ { "@type": "Question", "name": "What is a temp?", "acceptedAnswer": { "@type": "Answer", "text": "A temp is a short lived agent started for one job. The orchestrator or any lasting agent can start one, and the app can start one to answer an inbound message. It does the job, reports done and leaves." } },+ { "@type": "Question", "name": "Can one office mix Claude Code, Codex and other engines?", "acceptedAnswer": { "@type": "Answer", "text": "Yes. Each agent picks its own engine and model when you add it, and can have its own token cap. Claude Code and Codex use their own sign in; OpenCode, Crush, Pi and Qwen take a key." } },+ { "@type": "Question", "name": "Does the office keep working when I close my laptop?", "acceptedAnswer": { "@type": "Answer", "text": "No. Agents, schedules, Slack and webhooks all run on your own computer, so they stop when the app closes or the laptop sleeps." } },+ { "@type": "Question", "name": "Why did my agent's session clear itself?", "acceptedAnswer": { "@type": "Answer", "text": "That is a default in 0.5.6 that saves tokens. When an agent's work switches and its session has run for 60 minutes, the app clears the session, and a handoff file in the agent's folder carries the thread across." } }]}@@ -172,175 +171,117 @@<span>/</span><a href="/blog/topics/guides/">Guides</a></nav>- <h1>Your First Hour With Munder Difflin: Onboarding to a Pro Office</h1>- <p class="dek">Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.</p>+ <h1>Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office</h1>+ <p class="dek">Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.</p><div class="byline"><span class="av" aria-hidden="true">CG</span><strong>Chaitanya Giri</strong><span class="dot" aria-hidden="true">·</span><time datetime="2026-07-03">Jul 3, 2026</time><span class="dot" aria-hidden="true">·</span>- <span>13 min read</span>+ <span>10 min read</span><span class="tag kind">Guides</span></div></header>- <div class="hero-wrap wrap"><figure class="fig fig-hero"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/hero.png" alt="Michael, the orchestrator, in the Pro app after dispatching four tasks to his agents" loading="lazy" decoding="async" /></figure></div>+ <div class="hero-wrap wrap"><figure class="fig fig-hero"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/hero.png" alt="Title card on white. Large ink type reads Your First Hour With Munder Difflin 0.5.6, with the line From first sign in to a working office and a pill reading Onboarding guide, beside a blue stopwatch whose yellow ring has gone once all the way round as it rings." loading="lazy" decoding="async" /></figure></div><div class="post-wrap wrap"><div class="prose">- <p>Your first hour with <a href="https://harnessmd.com/download">Munder Difflin</a> 0.5.2 goes like this. Download the app, install one coding CLI, sign in and name your orchestrator. Then activate Pro and set the office up to work without you: a mixed team of agents, a pull request reviewer, a Slack responder, a morning email brief and a content agent on a schedule.</p>-<p>The app screenshots below are the real 0.5.2 Pro app, with personal details blurred. The order follows a new user from the download page to a working office.</p>-<h2 id="where-do-you-download-munder-difflin" tabindex="-1">Where do you download Munder Difflin? <a class="anchor" href="#where-do-you-download-munder-difflin" aria-hidden="true">#</a></h2>-<p>Download it from <a href="https://harnessmd.com/download">harnessmd.com/download</a>. You get there from <strong>Download free</strong> on <a href="https://munderdiffl.in/">munderdiffl.in</a>, or from <strong>Download</strong> in its top bar.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/a1-landing.png" alt="The top of munderdiffl.in with the Download free button" loading="lazy" decoding="async" /></figure>-<p>On 14 Sep 2026 the page listed version 0.5.2, published 9 Sep 2026, in three builds:</p>-<ul>-<li><strong>macOS:</strong> one build for Apple silicon and Intel, 246 MB.</li>-<li><strong>Windows:</strong> x64 installer, 132 MB.</li>-<li><strong>Linux:</strong> x86_64 AppImage, 175 MB.</li>-</ul>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/a3-download.png" alt="The Munder Difflin 0.5.2 download card with macOS, Windows and Linux builds" loading="lazy" decoding="async" /></figure>-<p>Open the file the usual way for your system: drag the app to Applications on a Mac, run the installer on Windows, or make the AppImage executable on Linux. <a href="https://harnessmd.com/download#whats-new">What changed in this version</a> is written out on the download page.</p>-<h2 id="what-do-you-need-installed-before-the-first-launch" tabindex="-1">What do you need installed before the first launch? <a class="anchor" href="#what-do-you-need-installed-before-the-first-launch" aria-hidden="true">#</a></h2>-<p>You need at least one coding CLI on your path, signed in with your own subscription or key. Munder Difflin supports twelve; this guide uses Claude Code, Codex and OpenCode. The commands below come from each vendor’s install docs, checked on 14 Sep 2026.</p>-<h3 id="claude-code" tabindex="-1">Claude Code <a class="anchor" href="#claude-code" aria-hidden="true">#</a></h3>-<pre class="language-bash"><code class="language-bash"><span class="token comment"># macOS, Linux, WSL</span>-<span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://claude.ai/install.sh <span class="token operator">|</span> <span class="token function">bash</span></code></pre>-<pre class="language-powershell"><code class="language-powershell"><span class="token comment"># Windows PowerShell</span>-<span class="token function">irm</span> https:<span class="token operator">/</span><span class="token operator">/</span>claude<span class="token punctuation">.</span>ai/install<span class="token punctuation">.</span>ps1 <span class="token punctuation">|</span> <span class="token function">iex</span></code></pre>-<p>Homebrew (<code>brew install --cask claude-code</code>) and WinGet (<code>winget install Anthropic.ClaudeCode</code>) work too. Run <code>claude</code> once in any folder and sign in with your Claude account.</p>-<h3 id="codex" tabindex="-1">Codex <a class="anchor" href="#codex" aria-hidden="true">#</a></h3>-<pre class="language-bash"><code class="language-bash"><span class="token comment"># macOS, Linux</span>-<span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://chatgpt.com/codex/install.sh <span class="token operator">|</span> <span class="token function">sh</span>-<span class="token comment"># or, with Node installed</span>-<span class="token function">npm</span> <span class="token function">install</span> <span class="token parameter variable">-g</span> @openai/codex</code></pre>-<p>Run <code>codex</code> and sign in with ChatGPT or an API key. Watch out for <code>npm install -g codex</code> without the <code>@openai/</code> scope: that is an unrelated package. <a href="/blog/how-to-install-codex-cli/">How to install Codex CLI</a> covers Windows and updates, and the <a href="/blog/how-to-install-and-use-munder-difflin/">Munder Difflin install guide</a> has the commands for all twelve CLIs.</p>-<h3 id="opencode" tabindex="-1">OpenCode <a class="anchor" href="#opencode" aria-hidden="true">#</a></h3>-<pre class="language-bash"><code class="language-bash"><span class="token comment"># macOS, Linux</span>-<span class="token function">curl</span> <span class="token parameter variable">-fsSL</span> https://opencode.ai/install <span class="token operator">|</span> <span class="token function">bash</span>-<span class="token comment"># or</span>-<span class="token function">npm</span> <span class="token function">install</span> <span class="token parameter variable">-g</span> opencode-ai</code></pre>-<p>On Windows, OpenCode’s docs list <code>choco install opencode</code> and <code>scoop install opencode</code>. Start <code>opencode</code> and run <code>/connect</code> to choose a model provider.</p>-<p>With Node installed, npm can put all three on your machine in one command. For the screenshot we installed them into a throwaway folder, then checked each with <code>--version</code>:</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/b1-terminal.png" alt="A terminal installing Claude Code, Codex and OpenCode with npm, then printing each version" loading="lazy" decoding="async" /></figure>-<p>If you skip this step, the app has a fallback. When you start an agent whose CLI is missing, 0.5.2 installs it in that agent’s own terminal, installing Node first if it has to, then relaunches the agent. You finish the CLI’s sign in right there. <strong>Settings, Prerequisites</strong> shows the status of each tool, with a <strong>Check now</strong> button. For memory search you also want <code>uv</code> and MemPalace, covered under memory below.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/b2-prerequisites.png" alt="Settings, Prerequisites in Munder Difflin listing agent engines, with Claude Code, Codex and OpenCode marked Found" loading="lazy" decoding="async" /></figure>-<h2 id="how-do-you-sign-in-on-the-first-launch" tabindex="-1">How do you sign in on the first launch? <a class="anchor" href="#how-do-you-sign-in-on-the-first-launch" aria-hidden="true">#</a></h2>-<p>You sign in through your browser with an email address; 0.5.2 asks for an account even on the free plan. The first screen asks <strong>How are you using Munder Difflin?</strong> Pick <strong>On my own</strong>, or <strong>With my team</strong> if an admin sent you an invite code.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c1-entry.png" alt="The first onboarding step, How are you using Munder Difflin, with On my own selected" loading="lazy" decoding="async" /></figure>-<p><strong>On my own</strong> leads to <strong>Sign in to continue</strong>, which opens your browser. If your browser does not come back to the app on its own, the screen lets you paste a code instead. New to Munder Difflin? The same sign in page on <a href="https://harnessmd.com/signup">harnessmd.com</a> creates the account.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c2-signin.png" alt="The Sign in to continue step with a box to paste the code from the browser" loading="lazy" decoding="async" /></figure>-<h2 id="how-do-you-set-up-your-workspace-and-orchestrator" tabindex="-1">How do you set up your workspace and orchestrator? <a class="anchor" href="#how-do-you-set-up-your-workspace-and-orchestrator" aria-hidden="true">#</a></h2>-<p>Three short steps follow the sign in, and the header counts all four for you.</p>+ <p>Your first hour of Munder Difflin onboarding goes like this: sign in, give your orchestrator one real job, hire a small team, set up work that starts without you, then learn the Pro screens. This guide is written for version 0.5.6 and follows the clock.</p>+<p><a href="https://harnessmd.com/download">Munder Difflin</a> is free and open source: a desktop app that runs a team of coding agents such as Claude Code, Codex and Gemini CLI on your own computer. Not installed yet? The <a href="/blog/how-to-install-and-use-munder-difflin/">install guide</a> covers the download, the engines and every fix. For ideas on what to hand your agents, <a href="/blog/ai-agent-use-cases-our-office/">every agent on our own floor</a> shows what ours do.</p>+<figure class="mg" data-scene="hour"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/hour.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A bus route map on white: one road runs left to right, bends round a hairpin past a pond and comes back, with six stops on it. Each stop has a minute mark and a name: Minute 0, sign in; Minute 10, a first job; Minute 20, a small team; Minute 35, work that starts without you; Minute 45, temps and tokens; Minute 50, the Pro screens. A small yellow bus drives from stop to stop, the road turns blue behind it, and each stop lights up yellow as the bus arrives. The bus leans through the hairpin and leaves two skid marks."><figcaption>The first hour as a route: sign in, a first job, a small team, work that starts without you, temps and tokens, then the Pro screens. Tap a stop to send the bus there.</figcaption></figure>+<h2 id="minute-0-what-do-you-need-and-how-do-you-sign-in" tabindex="-1">Minute 0: what do you need, and how do you sign in? <a class="anchor" href="#minute-0-what-do-you-need-and-how-do-you-sign-in" aria-hidden="true">#</a></h2>+<p>You need the app, one AI engine and an email address. The app needs nothing else installed, and it can install a missing engine for you later.</p>+<p>The first launch asks <strong>How are you using Munder Difflin?</strong> Pick <strong>On my own</strong>, then:</p><ol>-<li><strong>Name your workspace.</strong> A workspace name (HarnessAgents by default), a folder for the app’s own files, and whether you are technical or non technical. That last answer only changes how much jargon the app shows.</li>-<li><strong>Meet your orchestrator.</strong> Name the one agent that runs the others, then pick its engine and model. Each engine shows <strong>Installed</strong>, <strong>Installs on first start</strong> or <strong>Not installed</strong>. On Claude Code the default model is Opus 4.8 with a 1M context window, which suits the job: the orchestrator does the planning and delegating.</li>-<li><strong>Open the workspace.</strong> A last card shows the orchestrator and your first agent, with <strong>Open the workspace</strong> at the bottom.</li>+<li>Press <strong>Open browser to sign in</strong>. The browser does not open by itself.</li>+<li>Sign in in the browser. The screen moves on by itself; if it does not, paste the code into <strong>Code from the browser</strong> and press <strong>Use this code</strong>.</li>+<li><strong>Name your workspace</strong>, <strong>Meet your orchestrator</strong>, then <strong>Open the workspace</strong>.</li></ol>-<p>We call the orchestrator Michael in this guide. Call yours whatever you like.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c4-orchestrator.png" alt="The Meet your orchestrator step: Michael on Claude Code, marked Installed, with Opus 4.8 1M" loading="lazy" decoding="async" /></figure>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c5-engines.png" alt="The orchestrator engine list open, from Claude Code and Codex down to OpenCode, Pi, Copilot and Cursor" loading="lazy" decoding="async" /></figure>-<h2 id="how-do-you-get-pro" tabindex="-1">How do you get Pro? <a class="anchor" href="#how-do-you-get-pro" aria-hidden="true">#</a></h2>-<p>When you open the workspace without a licence, the app shows <strong>Go PRO on this machine</strong>. You have three choices:</p>+<p>Next comes the Pro sales screen, <strong>Run your agents like a company</strong>. Press <strong>Continue with the free version</strong> to reach the Classic view, or <strong>I have a licence key</strong> to turn on Pro. The steps through minute 45 work in both views; a few labels below are the Pro view’s.</p>+<h2 id="minute-10-how-do-you-give-your-orchestrator-a-first-job" tabindex="-1">Minute 10: how do you give your orchestrator a first job? <a class="anchor" href="#minute-10-how-do-you-give-your-orchestrator-a-first-job" aria-hidden="true">#</a></h2>+<p>Brief him like a capable new hire: the outcome, the folder, and what done looks like. Your orchestrator is the one agent you talk to. He is called Michael until you rename him.</p>+<p>A vague brief gets a vague job. Compare:</p><ul>-<li><strong>Continue with the free version</strong> takes you to the classic office, free, with everything local.</li>-<li><strong>Get PRO</strong> opens checkout in your browser. Pro runs on one machine at a time. Plans and prices are on the <a href="https://munderdiffl.in/#pricing">pricing page</a>.</li>-<li><strong>I have a licence key</strong> opens <strong>Enter your license key</strong>. Paste the key from your console at <a href="http://app.harnessmd.com">app.harnessmd.com</a>, press <strong>Activate</strong>, and it is bound to this machine.</li>+<li>Weak: <em>“clean up my project”</em></li>+<li>Better: <em>“In the folder notes-app, find every function with no test. Write the tests, run them, and give me a list of what failed and why. Do not change any function.”</em></li></ul>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c7-paywall.png" alt="The Go PRO on this machine screen listing what Pro adds, with Continue with the free version, I have a licence key and Get PRO" loading="lazy" decoding="async" /></figure>-<p>A key that is already active on another machine is refused: release it in your console, then activate it again. After that the app rechecks the licence in the background, and you do nothing.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/c8-license.png" alt="The Enter your license key screen with an empty key field and an Activate button" loading="lazy" decoding="async" /></figure>-<h2 id="what-does-a-set-up-pro-office-look-like" tabindex="-1">What does a set up Pro office look like? <a class="anchor" href="#what-does-a-set-up-pro-office-look-like" aria-hidden="true">#</a></h2>-<p>A Pro office is one window. The sidebar lists the orchestrator and your agents, and has a screen each for <strong>Agents</strong>, <strong>Tasks</strong>, <strong>Inbox</strong>, <strong>Automations</strong>, <strong>Memory</strong>, <strong>Capabilities</strong>, <strong>Stapler</strong> and <strong>Temps</strong> (short lived agents the orchestrator starts for one job).</p>-<p>The <strong>Agents</strong> screen puts the orchestrator on top, with today’s spend, the circuit breaker, open tasks and questions waiting on you. Below him, each agent has a card with its engine, model, current ticket and a live terminal preview.</p>-<p>You give work to the orchestrator in his terminal, the way you would brief a capable new hire: the outcome, the folder, and what done looks like. He turns it into tasks, hands them to agents and brings you only what needs your decision. How he decides is in <a href="/blog/how-the-god-orchestrator-works/">how the orchestrator routes work</a>.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/d2-agents.png" alt="The Pro Agents screen with Michael's budget and breaker cards above Jim, Dwight, Angela on Codex and Oscar on OpenCode" loading="lazy" decoding="async" /></figure>-<h2 id="what-can-your-office-do-in-its-first-hour" tabindex="-1">What can your office do in its first hour? <a class="anchor" href="#what-can-your-office-do-in-its-first-hour" aria-hidden="true">#</a></h2>-<p>Each of these six use cases takes a few minutes to set up.</p>-<h3 id="use-case-a-mixed-team-of-agents" tabindex="-1">Use case: a mixed team of agents <a class="anchor" href="#use-case-a-mixed-team-of-agents" aria-hidden="true">#</a></h3>-<p>Different models are good at different work, so give each agent the engine that fits its job. In <strong>Agents</strong>, press <strong>Add an agent</strong>. Under <strong>Providers</strong>, pick the provider (Claude Code, Codex, OpenCode or any of the others) and the model. <strong>Command</strong> then shows the exact command that starts the agent. Last comes the <strong>Token cap</strong>, the session budget before a restart: <strong>Workspace budget</strong>, <strong>200k</strong>, <strong>800k</strong>, <strong>2M</strong> or <strong>Other</strong>. When an agent passes its cap, the circuit breaker steers it, then constrains it, and stops it only if Hard stop is on.</p>-<p>One split to try: a Claude Code agent writing features, a Codex agent reviewing every diff, and an OpenCode agent on a cheaper model for docs and chores. OpenCode, Crush, Pi and Qwen read provider keys you save under <strong>Settings, Agents & Models</strong>; Claude Code and Codex use their own logins.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/d3-add-agent.png" alt="Add an agent: the Provider list, Model set to Sonnet 5, the exact command, and Token cap chips with 800k chosen" loading="lazy" decoding="async" /></figure>-<p>Sign each CLI in before you hire it. In the office we set up for these screenshots, the Codex account had no credits and the OpenCode key was invalid, so Angela and Oscar sat idle and Michael moved their tasks to Dwight, a Claude Code agent.</p>-<p>Every agent also gets its own room: the live terminal on one side, and its goal, engine, capabilities and token cap on the other.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/d4-agent.png" alt="Jim's agent room: a live Claude Code terminal reviewing a change, with an 800k token cap in force on the right" loading="lazy" decoding="async" /></figure>-<h3 id="use-case-an-agent-that-reviews-new-pull-requests" tabindex="-1">Use case: an agent that reviews new pull requests <a class="anchor" href="#use-case-an-agent-that-reviews-new-pull-requests" aria-hidden="true">#</a></h3>-<p>0.5.2 has no built in GitHub trigger, so use a schedule. In <strong>Automations</strong>, press <strong>New automation</strong>, then <strong>Schedule</strong>. Set the interval (15 minutes up to 24 hours), send it to your reviewer agent, and write the message: run <code>gh pr list</code> for the repo, review any pull request opened since the last sweep, and post the review as a comment. The agent needs the GitHub CLI signed in on your machine.</p>-<p>Want reviews the moment a pull request opens? A GitHub Action can post to a Munder Difflin webhook instead; see the webhook use case below.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/e1-pr-sweep.png" alt="A scheduled automation named PR review sweep, every 30 minutes, sent to Jim" loading="lazy" decoding="async" /></figure>-<h3 id="use-case-an-agent-that-answers-slack" tabindex="-1">Use case: an agent that answers Slack <a class="anchor" href="#use-case-an-agent-that-answers-slack" aria-hidden="true">#</a></h3>-<p>Your team mentions the bot in a Slack channel, and an agent does the work and replies in the thread. The basics:</p>-<ol>-<li>Create a Slack app at <a href="http://api.slack.com/apps">api.slack.com/apps</a> with the bot scopes <code>channels:history</code>, <code>groups:history</code>, <code>chat:write</code> and <code>files:read</code>, plus <code>channels:read</code> and <code>groups:read</code> so the app can list your channels. Install it and invite the bot to your channel.</li>-<li>In <strong>Settings, Connections</strong>, choose <strong>Who answers inbound messages</strong>. The orchestrator is the recommended default.</li>-<li>In the <strong>Slack</strong> card, paste the bot token, press <strong>Test connection</strong>, pick the channel, then press <strong>Turn on</strong>. The card starts on <strong>Check for messages</strong>, the recommended way in, which you can change under <strong>Advanced</strong>.</li>-</ol>-<p>The full walkthrough, with all three connection modes and the common failures, is <a href="/blog/connect-slack-to-munder-difflin/">How to Connect Slack to Munder Difflin 0.5.2</a>.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/h3-slack.png" alt="The Slack card in Settings, Connections with the bot token field, required scopes and the three ways Slack reaches the office" loading="lazy" decoding="async" /></figure>-<h3 id="use-case-a-morning-email-and-calendar-brief" tabindex="-1">Use case: a morning email and calendar brief <a class="anchor" href="#use-case-a-morning-email-and-calendar-brief" aria-hidden="true">#</a></h3>-<p>Claude Code has Gmail and Google Calendar connectors. Add them once at <a href="https://claude.ai/customize/connectors">claude.ai/customize/connectors</a>, and every Claude Code session signed in with that <a href="http://claude.ai">claude.ai</a> subscription can use them, which includes your Claude Code agents in Munder Difflin. <a href="https://code.claude.com/docs/en/mcp">Claude Code’s MCP docs</a> note that connectors do not load when an API key is the active login, and <code>/mcp</code> in the agent’s terminal shows whether they are there.</p>-<p>Then schedule the brief. <strong>New automation, Schedule</strong>, set <strong>When</strong> to <strong>Weekly</strong>, pick Monday to Friday at 08:30, send it to a Claude Code agent, and write: “Use the Gmail and Google Calendar connectors to summarise unread email and today’s meetings. Write it to <code>brief.md</code>.” Weekly times use your computer’s local clock.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/e2-morning-brief.png" alt="A weekly automation named Morning brief, Monday to Friday at 08:30, asking a Claude Code agent to use the Gmail and Google Calendar connectors" loading="lazy" decoding="async" /></figure>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/e7-connectors.png" alt="The /mcp list in a Claude Code agent's terminal showing claude.ai Gmail and Google Calendar connected" loading="lazy" decoding="async" /></figure>-<h3 id="use-case-a-content-agent-that-publishes-on-a-schedule" tabindex="-1">Use case: a content agent that publishes on a schedule <a class="anchor" href="#use-case-a-content-agent-that-publishes-on-a-schedule" aria-hidden="true">#</a></h3>-<p>This blog runs this way. An agent takes the next topic from a backlog every few hours, researches and writes a post, runs its checks, and opens a pull request that a person merges. For your own version, schedule a message to the orchestrator every 6 hours: “Write one blog post from the backlog, run the checks, open a pull request.” Keep the human merge; it is where mistakes get caught.</p>-<p>Every automation sits in one list, with filters for <strong>Scheduled</strong>, <strong>Context</strong> and <strong>Webhooks</strong>, and a <strong>History</strong> of what fired. The list starts with an <strong>Hourly ops standup</strong> that pings the orchestrator every hour (on by default, switched off in this screenshot), and context rules that compact or clear an agent’s terminal when its context fills. Turn off the standup if you do not want those tokens spent.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/e4-automations.png" alt="The Automations list with PR review sweep, Morning brief, Blog writer and the default automations" loading="lazy" decoding="async" /></figure>-<h3 id="use-case-run-a-smoke-test-when-a-deploy-finishes-webhooks" tabindex="-1">Use case: run a smoke test when a deploy finishes (webhooks) <a class="anchor" href="#use-case-run-a-smoke-test-when-a-deploy-finishes-webhooks" aria-hidden="true">#</a></h3>-<p>Webhooks let any tool that can send an HTTP request start work. In <strong>Automations</strong>, choose <strong>New automation, Webhook endpoint</strong>. The endpoint gets a public address once the webhook server is running and the tunnel picks it up. Each endpoint also has:</p>+<p>Your orchestrator turns the brief into tasks and works them, hands them to an agent you added, or starts a temp. Once there is one, click any agent to watch its terminal. To read what it wrote, press <strong>IDE</strong>: an editor opens on the agent’s folder, with <strong>Changes</strong> to show what was edited and a Markdown <strong>preview</strong> for reports.</p>+<figure class="mg" data-scene="brief"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/brief.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Two notepads on white above a long toaster marked your orchestrator. The pad tagged Weak reads clean up my project; the pad tagged Better reads the outcome, the folder, what done looks like. A slip drops from the weak pad into the first slot, the toaster works, and one grey card with a question mark pops up, wobbles and keeps smoking a little. A slip drops from the better pad into the second slot, and three task cards pop up, hop apart and each one gets a tick in its box."><figcaption>A brief with the outcome, the folder and what done looks like turns into clear tasks. A vague one turns into a question. Tap a note to drop it again.</figcaption></figure>+<h2 id="minute-20-how-do-you-hire-a-small-team" tabindex="-1">Minute 20: how do you hire a small team? <a class="anchor" href="#minute-20-how-do-you-hire-a-small-team" aria-hidden="true">#</a></h2>+<p>Press <strong>Add an agent</strong>. Pick a face, an engine, a model and the folder it works in. Each agent can run on a different engine, and that is the point: put your best model on the hard work and a cheaper one on the chores.</p>+<p>One split to try on day one:</p><ul>-<li><strong>Secret:</strong> senders send it in the <code>x-md-webhook-secret</code> header, and <strong>Regenerate</strong> revokes every sender at once.</li>-<li><strong>Gate:</strong> the default, strict, asks you before anything reaches the workspace.</li>-<li><strong>Prompt sent with every request:</strong> a standing instruction that rides ahead of the caller’s own text. The screenshot shows a support desk example; for deploys, try “Run the smoke test and report what fails. Never deploy.”</li>-<li><strong>A security warning switch:</strong> it adds a short note that the message is data, not instructions.</li>-<li><strong>Payload schema:</strong> a request whose body does not match is refused.</li>+<li>a builder on <strong>Claude Code</strong>, writing the feature</li>+<li>a reviewer on <strong>Codex</strong>, reading every change the builder makes</li>+<li>a helper on a free engine such as Antigravity, for docs and tidying</li></ul>-<p>The sender posts JSON with a <code>message</code> field and the secret in the <code>x-md-webhook-secret</code> header:</p>-<pre class="language-bash"><code class="language-bash"><span class="token function">curl</span> <span class="token parameter variable">-X</span> POST <span class="token string">"<span class="token variable">$MD_WEBHOOK_URL</span>"</span> <span class="token punctuation">\</span>- <span class="token parameter variable">-H</span> <span class="token string">"x-md-webhook-secret: <span class="token variable">$MD_WEBHOOK_SECRET</span>"</span> <span class="token punctuation">\</span>- <span class="token parameter variable">-H</span> <span class="token string">"content-type: application/json"</span> <span class="token punctuation">\</span>- <span class="token parameter variable">-d</span> <span class="token string">'{"message":"Deploy finished, run the smoke test","title":"deploy"}'</span></code></pre>-<p>With the default strict gate, the request waits in <strong>Inbox, Outside</strong> until you approve it, and only then does the orchestrator get a task card. For an unattended smoke test, set the gate to <strong>allow all</strong>. Two limits to know: new endpoints start switched off, and the tunnel URL changes every time the app restarts, so update the sender after a restart.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/e5-webhook.png" alt="A webhook endpoint sheet: name, endpoint, secret, strict gate, standing prompt, security warning switch and payload schema" loading="lazy" decoding="async" /></figure>-<h2 id="which-pro-features-should-you-learn-next" tabindex="-1">Which Pro features should you learn next? <a class="anchor" href="#which-pro-features-should-you-learn-next" aria-hidden="true">#</a></h2>-<p>Learn the screens you will open every day: tasks, inbox, memory and capabilities. Then add the voice features and the Stapler.</p>-<h3 id="feature-tasks-board" tabindex="-1">Feature: Tasks board <a class="anchor" href="#feature-tasks-board" aria-hidden="true">#</a></h3>-<p><strong>Tasks</strong> shows every card in the office as a board, a list or cards, with filters for <strong>Asks me</strong>, <strong>Unassigned</strong> and <strong>Archived</strong>, plus search. <strong>New task</strong> opens the orchestrator’s dispatch box so work still goes through him. If you drag cards between columns, the moves wait as proposals until you press <strong>Send to Michael</strong>.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/f1-tasks.png" alt="The Pro Tasks board with cards in To do, Doing, Blocked and Done" loading="lazy" decoding="async" /></figure>-<h3 id="feature-inbox" tabindex="-1">Feature: Inbox <a class="anchor" href="#feature-inbox" aria-hidden="true">#</a></h3>-<p><strong>Inbox</strong> puts every conversation in one place. <strong>For you</strong> holds the questions waiting on you and the orchestrator’s messages. <strong>Your team</strong> has a row per agent, and <strong>Everyone</strong> shows all routed messages. <strong>Outside</strong> collects Slack threads and webhook requests, and held webhook requests are approved or rejected right there.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/f2-inbox.png" alt="The Pro Inbox with For you, Your team, Everyone and Outside, showing Dwight's done message" loading="lazy" decoding="async" /></figure>-<h3 id="feature-memory-search" tabindex="-1">Feature: Memory search <a class="anchor" href="#feature-memory-search" aria-hidden="true">#</a></h3>-<p><strong>Memory</strong> searches every agent’s notes at once. Type a query, and results come back in three groups, <strong>Tickets</strong>, <strong>Agents</strong> and <strong>Memories</strong>, which you can narrow with <strong>Whose memory</strong> and <strong>Kind</strong>. Each ticket links to the agents and notes connected to it, and you can send selected results to an agent. Exact text search works out of the box. Search by meaning needs <code>uv</code> and MemPalace (<code>uv tool install mempalace</code>); <strong>Search across memory</strong> under <strong>Settings, Memory & Knowledge</strong> is on by default.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/f4-memory.png" alt="Memory search results for test, grouped as Tickets, Agents and Memories, across everyone's memory" loading="lazy" decoding="async" /></figure>-<h3 id="feature-capabilities-mcp-servers-and-skills" tabindex="-1">Feature: Capabilities, MCP servers and skills <a class="anchor" href="#feature-capabilities-mcp-servers-and-skills" aria-hidden="true">#</a></h3>-<p><strong>Capabilities</strong> is a marketplace we curate and check against each release: skills, MCP servers, plugins and engines. <strong>Install</strong> puts a skill in place for your Claude Code agents. Plugins install from inside a Claude Code session: press <strong>Copy command</strong> and paste it into the agent’s terminal. An MCP server can be turned on for the whole workspace or limited with <strong>Choose agents</strong>. <strong>Who has what</strong> shows which agent holds which capability, and <strong>Grant a role bundle</strong> sets an agent up in one go, with eleven role bundles to choose from. MCP servers reach Claude Code agents only.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/f5-capabilities.png" alt="The Capabilities marketplace, a curated catalog with Skills, MCP, Engines and Plugins filters" loading="lazy" decoding="async" /></figure>-<h3 id="feature-talk-and-dictation" tabindex="-1">Feature: Talk and dictation <a class="anchor" href="#feature-talk-and-dictation" aria-hidden="true">#</a></h3>-<p><strong>Talk</strong> is a live voice conversation with the orchestrator on OpenAI’s Realtime API. Save an OpenAI API key under <strong>Settings, Voice</strong>, then press <strong>Talk</strong> on the orchestrator’s screen. OpenAI bills the voice usage to your key, and the session disconnects after 3 idle minutes by default.</p>-<p><strong>Free Flow</strong> is dictation on Groq’s Whisper. It is on by default in the same Voice settings; save a Groq API key and pick a model, then click the mic above Send, or hold Option while you look at an agent’s terminal.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/f8-voice.png" alt="Settings, Voice: Free Flow dictation with a Groq API key field, and Realtime Michael with an OpenAI API key field" loading="lazy" decoding="async" /></figure>-<h3 id="feature-stapler" tabindex="-1">Feature: Stapler <a class="anchor" href="#feature-stapler" aria-hidden="true">#</a></h3>-<p>The <strong>Stapler</strong> is a small floating window that stays above every app, full screen ones included, and sends things to the orchestrator from anywhere. Turn on <strong>Show the stapler</strong> in the Stapler screen, then click it:</p>+<p>In the Pro view, each agent also has a <strong>Token cap: the session budget before a restart</strong>. Leave it on <strong>Workspace budget</strong> unless one agent needs its own. Sign every engine in before you hire on it, or the new agent cannot start; a Claude Code, Codex or Cursor agent stops at <strong>Sign in needed</strong>.</p>+<p>Our own floor has eleven kinds of seat: an orchestrator, mission control, a community lead, growth, motion graphics, research, a founder’s office, a scrum master, an engineering lead, builders, and a designer. You do not need eleven. You need a reviewer.</p>+<figure class="mg" data-scene="team"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/team.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three dotted outlines of office chairs stand in a row on white under a yellow button reading Add an agent, with the roles builder, reviewer and helper below them. A pointer presses the button three times; each press rolls in a chair, and an engine tag appears under it: on Claude Code, on Codex, on a free engine. The third chair arrives still spinning. A sheet of paper leaves the builder's chair, crosses to the reviewer's chair, is read line by line, gets a tick and comes back. Then the helper's heap of papers squares itself into a neat stack."><figcaption>A first team of three: a builder on Claude Code, a reviewer on Codex that reads every change, and a helper on a free engine for docs and tidying. Tap the builder to send another change for review.</figcaption></figure>+<h2 id="minute-35-how-do-you-set-up-work-that-starts-without-you" tabindex="-1">Minute 35: how do you set up work that starts without you? <a class="anchor" href="#minute-35-how-do-you-set-up-work-that-starts-without-you" aria-hidden="true">#</a></h2>+<p>With an automation. In the Pro view open <strong>Automations</strong> and press <strong>New automation</strong>; in the Classic view the same things live under Triggers. Two kinds start there, a schedule and a webhook, and Slack is a third way in.</p>+<p><strong>A schedule</strong> sends a message to an agent on a clock. Choose <strong>Schedule</strong>, then fill in four things: a <strong>Label</strong>, <strong>When</strong> (an <strong>Interval</strong> or <strong>Weekly</strong>), <strong>Sends to</strong>, and the <strong>Message</strong>. The message is what the agent should do when it fires. A good first one: every weekday morning, ask an agent to check something and write you three lines.</p>+<figure class="mg" data-scene="schedule"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/schedule.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. A card on white titled Automations with a yellow New automation button; on the right, a horizon with an empty sky and an office chair. A pointer presses New automation, picks Schedule over Webhook endpoint, and four fields fill in: Label, morning check; When, Weekly rather than Interval; Sends to, an agent; Message, check something and write you three lines. The pointer leaves. A sun climbs over the horizon, the card fires by itself, and a paper slip flies to the chair and gets a tick. The pointer peeks back in to check, then leaves again."><figcaption>A schedule has four parts: a Label, When (an Interval or Weekly), Sends to, and the Message the agent gets every time it fires. Tap the card to fire it again.</figcaption></figure>+<p><strong>A webhook</strong> lets any tool that can send a web request start work. Choose <strong>Webhook endpoint</strong>. A new endpoint starts switched off: choose how much you trust the caller, copy its address and its secret into the tool that will send, then turn it on. The sender must hold the secret.</p>+<p><strong>Slack</strong> lets people send work from a channel and read the answer in the same thread. It is under <strong>Settings</strong>, then <strong>Connections</strong>. In 0.5.6 one bot can listen to <strong>Every channel the bot is in</strong>, and a direct message to the bot is always answered. For direct messages, add <code>im:history</code>, <code>im:read</code>, <code>mpim:history</code> and <code>mpim:read</code> on your Slack app’s <strong>OAuth & Permissions</strong> page at <a href="http://api.slack.com">api.slack.com</a>, then follow the app’s own <strong>How to get these</strong> steps for the events. You choose the agent under <strong>Who answers inbound messages</strong>. If Slack fails to start, it now comes back by itself. The <a href="/blog/connect-slack-to-munder-difflin/">Slack guide</a> has the full setup.</p>+<figure class="mg" data-scene="slack"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/slack.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a Connections card under Settings with a Slack switch and a box for Every channel the bot is in; below it, three chat cards labelled a channel, another channel and a direct message. A pointer switches Slack on and ticks the box, and an office chair rolls into the empty seat under Who answers inbound messages. A string runs from each card to the chair. Each card posts a message, a note travels down its string to the chair, and a reply comes back and lands under that message as a thread. Then Slack switches itself off, the strings go slack, and it switches itself back on."><figcaption>In 0.5.6 one bot can hear every channel it is in, a direct message is always answered, and the reply comes back in the same thread. Tap a card to send a message.</figcaption></figure>+<h2 id="minute-45-what-are-temps-and-how-do-you-keep-token-use-down" tabindex="-1">Minute 45: what are temps, and how do you keep token use down? <a class="anchor" href="#minute-45-what-are-temps-and-how-do-you-keep-token-use-down" aria-hidden="true">#</a></h2>+<p>A temp is an agent for one job. It gets an objective, does the work, reports done and leaves. There is no button for one. You ask: <em>“Michael, start a temp to rename every file in this folder to lower case.”</em></p>+<p>Three things changed in 0.5.6:</p><ul>-<li><strong>Screenshot:</strong> drag a box, add a note (“What should Michael do with this?”), send. The first capture makes macOS ask for screen recording permission.</li>-<li><strong>Record message:</strong> speak, read the transcript, send.</li>-<li><strong>Record meeting:</strong> records your microphone and writes a transcript you can send from <strong>Meeting transcripts</strong>.</li>-<li><strong>Leave invisible:</strong> hides the Stapler from screen sharing.</li>+<li><strong>Any lasting agent can start a temp</strong>, not only the orchestrator. A temp cannot start a temp.</li>+<li><strong>A temp can answer inbound messages.</strong> Wherever you pick who answers (Slack, a webhook, a schedule’s <strong>Sends to</strong>, the Stapler’s <strong>Send captures to</strong>), the list now offers <strong>A temp</strong> and a temp that is a clone of one of your agents.</li>+<li><strong>You set the limits.</strong> In the Pro view, on the orchestrator’s <strong>Budget & breaker</strong> tab, <strong>Temps per agent</strong> runs from 0 to 4 and <strong>Max temps at once</strong> caps how many temps run in the whole office at the same time. At 0, agents cannot start temps.</li></ul>-<p>Recording needs the Groq key: until one is saved, <strong>Record message</strong> and <strong>Record meeting</strong> stay greyed out on the ring. Under <strong>Actions, Send captures to</strong>, you can point captures at any agent instead of the orchestrator.</p>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/g1-stapler.png" alt="The Stapler screen with the stapler on screen and its face, shape, expression, size and colour settings" loading="lazy" decoding="async" /></figure>-<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/g3-ring.png" alt="The floating Stapler with its action ring open" loading="lazy" decoding="async" /></figure>-<h2 id="what-should-you-change-on-day-one" tabindex="-1">What should you change on day one? <a class="anchor" href="#what-should-you-change-on-day-one" aria-hidden="true">#</a></h2>-<p>Check four defaults before you leave the office alone. Setup turns them on or off for you:</p>+<figure class="mg" data-scene="temp"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/temp.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. Three office chairs stand on white over the words your lasting agents, beside the dotted outline of a folding chair. A message card reading one job waits at the top left, then flies across while a yellow folding chair slides in folded, half opens, snaps shut, then opens properly and gets a tag reading A temp. The message lands on its seat, a done card pops up and flies to the first office chair, and the folding chair folds flat and slides away. The three office chairs never move."><figcaption>A temp answers one message or does one job, reports done and leaves; your lasting agents keep their seats. Tap the message to send another.</figcaption></figure>+<p>Tokens are the other limit. Long sessions fill an agent’s context, and a full context is slow and uses up your plan’s limits faster. 0.5.6 ships with defaults that deal with it:</p><ul>-<li><strong>Auto mode is on.</strong> Agents act without stopping to ask permission. That is what lets them work unattended, and it is also why they belong in folders you are happy for them to change. The switch is <strong>Auto mode</strong> in the orchestrator’s <strong>Budget & breaker</strong> tab.</li>-<li><strong>The orchestrator cannot start agents on his own</strong> until you turn on <strong>Can start agents on its own</strong> in his <strong>Configuration</strong> tab. Turn it on if you want temps.</li>-<li><strong>The Hourly ops standup is on.</strong> Keep it if you like the check ins; otherwise switch it off in Automations.</li>-<li><strong>New Claude Code agents default to Fable 5.</strong> The orchestrator starts on Opus 4.8. As of September 2026, <a href="https://support.claude.com/en/articles/15424964-claude-fable-models-on-your-plan">Fable runs on usage credits</a> on Claude Pro; on Max it can use up to half of your weekly limit at no extra cost, then usage credits. Pick another default under <strong>Settings, Agents & Models</strong> if you run many agents.</li>+<li><strong>Compact when context fills</strong> summarises a long session and carries the summary forward.</li>+<li><strong>Clear on a switch of work</strong> clears the session when an agent moves to different work and the session has run for 60 minutes. A file called <code>handoff.md</code> in the agent’s folder carries the thread across, so the next session knows where the last one stopped.</li>+<li><strong>No hourly standup.</strong> A new install no longer pings the orchestrator every hour.</li></ul>-<p>That is the hour: download, one CLI, sign in, Pro, a mixed team, and five jobs that keep running while you do something else. Once it runs, <a href="/blog/build-your-personal-agi-on-your-computer/">build your personal AGI</a> turns that office into a team that briefs you, sweeps pull requests, remembers you and answers from Slack. <a href="https://harnessmd.com/download">Download Munder Difflin</a>, and if the hour earns it, <a href="https://github.com/chaitanyagiri/munder-difflin">a GitHub star</a> helps other people find it.</p>+<p>The first two are rules in <strong>Automations</strong> under <strong>Context</strong> (in the Classic view, the Triggers tab). Leave them on.</p>+<figure class="mg" data-scene="tokens"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/tokens.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a tall glass on a coaster stands beside a card titled Context under Automations, with two rules switched on: Compact when context fills, and Clear on a switch of work. A pill below the card reads Leave them on. Paper slips drop into the glass until the pile nears a dashed line. The first rule lights up, a press with a yellow knob swings over and pushes down, and the slips squash into one slip labelled summary. A few more slips land on top. Then the second rule lights up: the glass turns upside down and is shaken, the slips fall out, the summary clings on until a second shake, and the glass turns back. A yellow note labelled with the handoff file's name drops into the empty glass and stays."><figcaption>Two defaults in 0.5.6 keep sessions small: Compact when context fills summarises a long session, and Clear on a switch of work clears it and leaves a handoff file to carry the thread. Tap the glass to fill it again.</figcaption></figure>+<h2 id="minute-50-what-does-the-pro-view-add" tabindex="-1">Minute 50: what does the Pro view add? <a class="anchor" href="#minute-50-what-does-the-pro-view-add" aria-hidden="true">#</a></h2>+<p>One window with a sidebar, and a screen for each part of the office. The switch in the title bar moves between <strong>Classic</strong> and <strong>PRO</strong>; without a licence it reads <strong>TRY PRO</strong>.</p>+<figure class="mg" data-scene="sidebar"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/sidebar.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. An app window on white with a switch in the title bar reading Classic and PRO, with PRO selected, and a sidebar of six rows: Tasks, Inbox, Automations, Memory, Capabilities, Stapler. A yellow highlight walks down the rows, and as each one lights, its one line note writes itself into the pane beside it: every card in the office; one conversation per agent; Scheduled, Context and Webhooks; Search every agent's memory; what each agent can reach; a small window that floats above every app. When the highlight reaches Stapler, that row lifts off the sidebar and floats above it with a small shadow under it."><figcaption>The Pro view is one window with a sidebar row for Tasks, Inbox, Automations, Memory, Capabilities and the Stapler. Tap a row to see what it is for.</figcaption></figure>+<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/pro-sidebar-056.png" alt="Screenshot of the Munder Difflin Pro sidebar in light mode: the app name and PRO plan, then rows for Tasks, Inbox, Automations, Memory, Capabilities and Stapler." loading="lazy" decoding="async" /><figcaption>The real Pro sidebar: Tasks, Inbox, Automations, Memory, Capabilities and Stapler. From a preview build of 0.5.6 with sample data.</figcaption></figure>+<p><strong>Tasks</strong> is every card in the office, as a <strong>Board</strong>, a <strong>List</strong> or <strong>Cards</strong>. The habit that pays: start the day on the <strong>Asks me</strong> filter, which shows only the cards waiting on you. (It is a filter in Tasks; <strong>Ask me</strong> in the Inbox, below, is where you answer.) You can drag cards between columns. Nothing changes for your team until you send your edits, and then your orchestrator makes the change.</p>+<p><strong>Inbox</strong> holds one conversation per agent, plus <strong>Ask me</strong>, <strong>Slack</strong> and <strong>Webhooks</strong>. <strong>Ask me</strong> is the list of questions waiting on a person. Answer in one line; answering unblocks the task. This is the screen that keeps an office moving while you do something else.</p>+<figure class="fig fig-inline"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/ask-me-056.png" alt="Screenshot of an agent's Inbox in Munder Difflin, light mode: an Ask me list with three questions, each with a task key, a time and an Answer button, above the message box with Send and Steer." loading="lazy" decoding="async" /><figcaption>Ask me in an agent's Inbox: each question waits with an Answer button. From a preview build of 0.5.6 with sample data.</figcaption></figure>+<figure class="mg" data-scene="askme"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/askme.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a large card with an Ask me chip and the word Inbox shows a question as grey lines with a lilac question mark, an answer field reading Answer in one line, a blue send button and a chip reading waiting on you. Beside it, three smaller question cards wait in a column under the words waiting on a person. A short answer types itself in as one line and is sent; the question mark turns into a mint tick and the chip becomes unblocks the task. The card slides away, the next question card grows into its place, and the queue shuffles up; the last card arrives a little too fast and bumps the one ahead. The new question gets its one line answer and its tick too."><figcaption>Ask me collects the questions waiting on a person. A short answer unblocks the task, and the card moves on. Tap the card to ask again.</figcaption></figure>+<p><strong>Automations</strong> is the list you met at minute 35, with filters for <strong>Scheduled</strong>, <strong>Context</strong> and <strong>Webhooks</strong>, Each schedule shows when it <strong>Last fired</strong>, each webhook lists its <strong>Recent</strong> arrivals, and <strong>History</strong> opens the Inbox, where webhook messages are answered.</p>+<p><strong>Memory</strong> lets you <strong>Search every agent’s memory</strong> at once and see the result as a <strong>Graph</strong> or a <strong>List</strong>. Use it before you brief: if an agent already solved this last week, the note is here. Narrow a search with <strong>Whose memory</strong> and <strong>Kind</strong>.</p>+<p><strong>Capabilities</strong> shows what each agent can reach: MCP servers, skills and tools.</p>+<p><strong>The Stapler</strong> is a small window that floats above every app. On the <strong>Stapler</strong> screen, turn on <strong>Show the stapler</strong>, then click it for its ring:</p>+<ul>+<li><strong>Screenshot</strong> captures a region of the screen and sends it, with a note, to the agent you pick under <strong>Send captures to</strong>.</li>+<li><strong>Record message</strong> takes a spoken message and shows you the transcript before it goes.</li>+<li><strong>Record meeting</strong> records the microphone into a folder of transcripts, one folder per meeting.</li>+<li><strong>Make invisible</strong> hides the Stapler from screen sharing and recordings.</li>+</ul>+<p>A fifth button, <strong>Computer use</strong>, reads <strong>Coming soon</strong> and cannot be turned on yet.</p>+<p>The habit: when you see a bug in any app, do not describe it. Staple it.</p>+<figure class="mg" data-scene="stapler"><img src="/blog/assets/media/your-first-hour-with-munder-difflin/stapler.png" width="1600" height="1200" loading="lazy" decoding="async" alt="Animation. On white, a plain app window made of grey bars, with a setting above it reading Show the stapler and a chip reading an agent. The setting switches on and a round stapler button pops up and floats over the corner of the app window. A pointer clicks it and a ring of four actions opens around it: Screenshot, Record message, Record meeting, Make invisible. Screenshot is picked, a crosshair drags a dashed box over part of the app, and the region lifts out as a capture. A note field reading your note slides in under it and a short line types itself in. The stapler's arm comes down and a staple joins the note to the capture, then the pair flies to the agent chip, which gets a tick. The ring opens again and stays."><figcaption>The Stapler floats above every app. Its ring offers Screenshot, Record message, Record meeting and Make invisible, plus Computer use, coming soon, and a capture goes to an agent with your note. Tap the stapler to open the ring.</figcaption></figure>+<h2 id="minute-58-what-about-accounts-and-teams" tabindex="-1">Minute 58: what about accounts and teams? <a class="anchor" href="#minute-58-what-about-accounts-and-teams" aria-hidden="true">#</a></h2>+<p>Your account sits in the profile menu. <strong>Sign out</strong> asks you to press again, so you cannot do it by accident. <strong>Switch to Teams</strong> opens one sheet for joining a team with an invite code or generating one.</p>+<h2 id="what-trips-people-up-in-the-first-hour" tabindex="-1">What trips people up in the first hour? <a class="anchor" href="#what-trips-people-up-in-the-first-hour" aria-hidden="true">#</a></h2>+<p>Eight things, all normal.</p>+<ol>+<li><strong>The paywall comes first.</strong> Press <strong>Continue with the free version</strong> to reach the app.</li>+<li><strong>The sign in browser does not open by itself.</strong> Press the button and sign in there. If the screen does not move on by itself, paste the code.</li>+<li><strong>“Installs on first start” still needs you.</strong> The engine installs when you press <strong>Install</strong>.</li>+<li><strong>An agent reads “looping”.</strong> The circuit breaker warns an agent that repeats itself or spends fast. That is the app working, not a crash.</li>+<li><strong>A session cleared itself.</strong> See the token rules above. <code>handoff.md</code> has the thread.</li>+<li><strong>Auto mode decides how much agents ask.</strong> On, agents act on their own and ask only when blocked. Off, they ask before each tool call. In the Pro view it is on the orchestrator’s <strong>Budget & breaker</strong> tab; in the Classic view, <strong>Settings</strong>, then <strong>Autonomy & Budgets</strong>. It applies when an agent next starts.</li>+<li><strong>Many agents on 8 GB of memory run slowly.</strong> Run fewer at once.</li>+<li><strong>On Linux with Wayland, dictation into other apps is refused.</strong> It needs X11.</li>+</ol>+<h2 id="what-should-you-do-after-the-first-hour" tabindex="-1">What should you do after the first hour? <a class="anchor" href="#what-should-you-do-after-the-first-hour" aria-hidden="true">#</a></h2>+<p>Add one seat at a time. A reviewer first, then one scheduled job, then a scout. <a href="/blog/ai-agent-use-cases-our-office/">Every agent on our own floor</a> has starter messages for four seats. <a href="/blog/how-the-god-orchestrator-works/">How the orchestrator routes work</a> explains what your orchestrator does with your briefs, and <a href="/blog/build-your-personal-agi-on-your-computer/">build your personal AGI</a> turns the office into something that briefs you each morning.</p>+<link rel="stylesheet" href="/blog/assets/media/your-first-hour-with-munder-difflin/motion.css"><script defer src="/blog/assets/media/your-first-hour-with-munder-difflin/motion.js"></script>@@ -349,32 +290,32 @@<div class="qa"><p class="q">How long does it take to get Munder Difflin running?</p>- <p class="a">About ten minutes if Claude Code, Codex or OpenCode is already installed and signed in: download, sign in, answer four setup steps. Allow the rest of the hour to hire a small team and set up your first automations.</p>+ <p class="a">Setup is four short steps when an AI engine is already installed and signed in: download, sign in, name your workspace and your orchestrator. The rest of the hour goes on a first job, a small team and the first work that starts without you.</p></div><div class="qa">- <p class="q">Do I need an account to use Munder Difflin?</p>- <p class="a">Yes, in 0.5.2. The first launch asks you to sign in with an email address in your browser, on the free plan as well. Pro then needs a licence key on top of that account.</p>+ <p class="q">Do I need Pro for this first hour?</p>+ <p class="a">No. Agents, schedules, the context rules, webhooks, Slack and temps are in both views. The Pro view adds one window with screens for Tasks, Inbox, Automations, Memory and Capabilities, plus the Stapler, a small window above every app.</p></div><div class="qa">- <p class="q">What does Pro add over the free version?</p>- <p class="a">The free version is the classic office. Pro adds a single window workspace with screens for the orchestrator, agents, agent rooms, tasks, inbox, automations, memory, capabilities and temps, plus the Stapler. It runs on one machine at a time; the pricing page has the rest.</p>+ <p class="q">What is a temp?</p>+ <p class="a">A temp is a short lived agent started for one job. The orchestrator or any lasting agent can start one, and the app can start one to answer an inbound message. It does the job, reports done and leaves.</p></div><div class="qa">- <p class="q">Can one Pro office mix Claude Code, Codex and OpenCode agents?</p>- <p class="a">Yes. Each agent picks its own provider and model when you add it, and can have its own token cap. Claude Code and Codex use their own logins; OpenCode can use a provider key you save under Settings.</p>+ <p class="q">Can one office mix Claude Code, Codex and other engines?</p>+ <p class="a">Yes. Each agent picks its own engine and model when you add it, and can have its own token cap. Claude Code and Codex use their own sign in; OpenCode, Crush, Pi and Qwen take a key.</p></div><div class="qa"><p class="q">Does the office keep working when I close my laptop?</p>- <p class="a">No. Agents, automations, Slack and webhooks all run on your machine, so they stop when the app closes or the laptop sleeps. A schedule missed while the laptop slept runs once when the app wakes; a weekly one only if that is within six hours of its slot.</p>+ <p class="a">No. Agents, schedules, Slack and webhooks all run on your own computer, so they stop when the app closes or the laptop sleeps.</p></div><div class="qa">- <p class="q">Is the Stapler free?</p>- <p class="a">No. The Stapler is Pro only, and it is off until you turn on **Show the stapler** on the Stapler screen.</p>+ <p class="q">Why did my agent's session clear itself?</p>+ <p class="a">That is a default in 0.5.6 that saves tokens. When an agent's work switches and its session has run for 60 minutes, the app clears the session, and a handoff file in the agent's folder carries the thread across.</p></div></section>@@ -384,7 +325,7 @@<hr /><div class="share"><span class="lbl">Tags</span>- <a class="tag" href="/blog/tags/guides/">Guides</a><a class="tag" href="/blog/tags/onboarding/">Onboarding</a><a class="tag" href="/blog/tags/getting-started/">Getting Started</a><a class="tag" href="/blog/tags/multi-agent/">Multi-Agent</a><a class="tag" href="/blog/tags/local-first/">Local-First</a>+ <a class="tag" href="/blog/tags/guides/">Guides</a><a class="tag" href="/blog/tags/onboarding/">Onboarding</a><a class="tag" href="/blog/tags/getting-started/">Getting Started</a><a class="tag" href="/blog/tags/multi-agent/">Multi-Agent</a></div></div>@@ -393,7 +334,7 @@<aside class="toc-rail" aria-label="Table of contents"><div class="tt">On this page</div><ol>- <li class="lvl-2"><a href="#where-do-you-download-munder-difflin">Where do you download Munder Difflin?</a></li><li class="lvl-2"><a href="#what-do-you-need-installed-before-the-first-launch">What do you need installed before the first launch?</a></li><li class="lvl-3"><a href="#claude-code">Claude Code</a></li><li class="lvl-3"><a href="#codex">Codex</a></li><li class="lvl-3"><a href="#opencode">OpenCode</a></li><li class="lvl-2"><a href="#how-do-you-sign-in-on-the-first-launch">How do you sign in on the first launch?</a></li><li class="lvl-2"><a href="#how-do-you-set-up-your-workspace-and-orchestrator">How do you set up your workspace and orchestrator?</a></li><li class="lvl-2"><a href="#how-do-you-get-pro">How do you get Pro?</a></li><li class="lvl-2"><a href="#what-does-a-set-up-pro-office-look-like">What does a set up Pro office look like?</a></li><li class="lvl-2"><a href="#what-can-your-office-do-in-its-first-hour">What can your office do in its first hour?</a></li><li class="lvl-3"><a href="#use-case-a-mixed-team-of-agents">Use case: a mixed team of agents</a></li><li class="lvl-3"><a href="#use-case-an-agent-that-reviews-new-pull-requests">Use case: an agent that reviews new pull requests</a></li><li class="lvl-3"><a href="#use-case-an-agent-that-answers-slack">Use case: an agent that answers Slack</a></li><li class="lvl-3"><a href="#use-case-a-morning-email-and-calendar-brief">Use case: a morning email and calendar brief</a></li><li class="lvl-3"><a href="#use-case-a-content-agent-that-publishes-on-a-schedule">Use case: a content agent that publishes on a schedule</a></li><li class="lvl-3"><a href="#use-case-run-a-smoke-test-when-a-deploy-finishes-webhooks">Use case: run a smoke test when a deploy finishes (webhooks)</a></li><li class="lvl-2"><a href="#which-pro-features-should-you-learn-next">Which Pro features should you learn next?</a></li><li class="lvl-3"><a href="#feature-tasks-board">Feature: Tasks board</a></li><li class="lvl-3"><a href="#feature-inbox">Feature: Inbox</a></li><li class="lvl-3"><a href="#feature-memory-search">Feature: Memory search</a></li><li class="lvl-3"><a href="#feature-capabilities-mcp-servers-and-skills">Feature: Capabilities, MCP servers and skills</a></li><li class="lvl-3"><a href="#feature-talk-and-dictation">Feature: Talk and dictation</a></li><li class="lvl-3"><a href="#feature-stapler">Feature: Stapler</a></li><li class="lvl-2"><a href="#what-should-you-change-on-day-one">What should you change on day one?</a></li>+ <li class="lvl-2"><a href="#minute-0-what-do-you-need-and-how-do-you-sign-in">Minute 0: what do you need, and how do you sign in?</a></li><li class="lvl-2"><a href="#minute-10-how-do-you-give-your-orchestrator-a-first-job">Minute 10: how do you give your orchestrator a first job?</a></li><li class="lvl-2"><a href="#minute-20-how-do-you-hire-a-small-team">Minute 20: how do you hire a small team?</a></li><li class="lvl-2"><a href="#minute-35-how-do-you-set-up-work-that-starts-without-you">Minute 35: how do you set up work that starts without you?</a></li><li class="lvl-2"><a href="#minute-45-what-are-temps-and-how-do-you-keep-token-use-down">Minute 45: what are temps, and how do you keep token use down?</a></li><li class="lvl-2"><a href="#minute-50-what-does-the-pro-view-add">Minute 50: what does the Pro view add?</a></li><li class="lvl-2"><a href="#minute-58-what-about-accounts-and-teams">Minute 58: what about accounts and teams?</a></li><li class="lvl-2"><a href="#what-trips-people-up-in-the-first-hour">What trips people up in the first hour?</a></li><li class="lvl-2"><a href="#what-should-you-do-after-the-first-hour">What should you do after the first hour?</a></li></ol></aside>@@ -407,50 +348,50 @@<h2>Related in Guides</h2><div class="post-grid"><article class="card t-guides">- <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/ai-agent-use-cases-our-office/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/ai-agent-use-cases-our-office/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-13">Oct 13, 2026</time>+ <time datetime="2026-10-14">Oct 14, 2026</time><span class="dot" aria-hidden="true"></span>- <span>4 min</span>+ <span>8 min</span><span class="pin">Pinned</span></div>- <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>- <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p>+ <h3><a href="/blog/ai-agent-use-cases-our-office/">AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does</a></h3>+ <p class="dek">We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a>+ <a class="more" href="/blog/ai-agent-use-cases-our-office/" aria-label="Read: AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/gemini-4-argon-pricing/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/gemini-4-argon-pricing/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta"><time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span><span>4 min</span></div>- <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>- <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p>+ <h3><a href="/blog/gemini-4-argon-pricing/">Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives</a></h3>+ <p class="dek">Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a>+ <a class="more" href="/blog/gemini-4-argon-pricing/" aria-label="Read: Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives">Read →</a></div></div></article><article class="card t-guides">- <a class="thumb t-guides" href="/blog/asd-ste100-prompting/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/asd-ste100-prompting/hero.png" alt="" loading="lazy" decoding="async" /></a>+ <a class="thumb t-guides" href="/blog/how-to-use-gemini-4-argon/" aria-hidden="true" tabindex="-1"><img src="/blog/assets/media/how-to-use-gemini-4-argon/hero.png" alt="" loading="lazy" decoding="async" /></a><div class="body"><div class="meta">- <time datetime="2026-10-05">Oct 5, 2026</time>+ <time datetime="2026-10-13">Oct 13, 2026</time><span class="dot" aria-hidden="true"></span>- <span>5 min</span>+ <span>4 min</span></div>- <h3><a href="/blog/asd-ste100-prompting/">ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers</a></h3>- <p class="dek">Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.</p>+ <h3><a href="/blog/how-to-use-gemini-4-argon/">How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them</a></h3>+ <p class="dek">How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.</p><div class="foot"><span class="tag kind">Guides</span>- <a class="more" href="/blog/asd-ste100-prompting/" aria-label="Read: ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers">Read →</a>+ <a class="more" href="/blog/how-to-use-gemini-4-argon/" aria-label="Read: How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them">Read →</a></div></div></article>diff --git a/docs/llms.txt b/docs/llms.txtindex d94229e8..31d3272c 100644--- a/docs/llms.txt+++ b/docs/llms.txt@@ -43,7 +43,60 @@<!-- BLOG:START generated by blog/scripts/postbuild.mjs -->## Blog-212 posts, grouped by topic, newest first in each topic. Topic hubs: https://munderdiffl.in/blog/topics/+213 posts, grouped by topic, newest first in each topic. Topic hubs: https://munderdiffl.in/blog/topics/++### guides++- [How to Install Munder Difflin 0.5.6: A Beginner's Guide From Zero to a Working Office](https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/): Install Munder Difflin 0.5.6 on macOS, Windows or Linux: what you need first, which AI engine to pick, the sign in, the setup steps and a fix for each failure.+- [Your First Hour With Munder Difflin 0.5.6: Onboarding From First Sign In to a Working Office](https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/): Your first hour with Munder Difflin 0.5.6: sign in, brief your orchestrator, hire a team, schedule work, connect Slack, start temps, then tour the Pro screens.+- [AI Agent Use Cases From a Real Company: Every Agent on Our Floor and the Job It Does](https://munderdiffl.in/blog/ai-agent-use-cases-our-office/): We run our company on Munder Difflin. Every seat on our floor: what each AI agent does, how its work starts, what comes out, and how to copy it.+- [How to Get the Most Out of Your Claude Code Max Plan](https://munderdiffl.in/blog/claude-code-max-plan-tips/): Stretch a Claude Code Max plan: model choice, context hygiene, prompt caching, subagents and unattended runs, checked against Anthropic's own pages.+- [How to Make the Most of Your Codex Max Plan](https://munderdiffl.in/blog/codex-max-plan-tips/): There is no Codex Max plan. What ChatGPT Pro 5x and 20x give you in Codex, how the weekly limit works, and how to stretch it.+- [Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives](https://munderdiffl.in/blog/gemini-4-argon-pricing/): Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.+- [How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them](https://munderdiffl.in/blog/how-to-use-gemini-4-argon/): How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.+- [ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers](https://munderdiffl.in/blog/asd-ste100-prompting/): Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.+- [Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)](https://munderdiffl.in/blog/codex-pricing/): Codex pricing checked 5 Oct 2026: which ChatGPT plans include Codex, US prices from $0 to $500, how usage limits are counted, credits and API key rates.+- [How to Use Claude Code: Setup, First Task and the Commands That Matter](https://munderdiffl.in/blog/how-to-use-claude-code/): How to use Claude Code, checked 5 Oct 2026: install it, sign in, run a first task, then the commands, shortcuts and habits a beginner needs.+- [Build Your Personal AGI on Your Computer Using Claude Code and Munder Difflin](https://munderdiffl.in/blog/build-your-personal-agi-on-your-computer/): Turn Claude Code into a personal AGI on your own computer: agents with memory, schedules, Gmail, Calendar, voice and Slack, set up in Munder Difflin 0.5.2.+- [How to Install Gemini CLI, and Can It Use Gemini 4 Yet?](https://munderdiffl.in/blog/how-to-install-gemini-cli/): Install Gemini CLI 0.62.0 with one npm command, see which sign ins still work after 18 Jun 2026, and why Gemini 4 Argon isn't in it yet.+- [How to Connect Slack to Munder Difflin 0.5.2](https://munderdiffl.in/blog/connect-slack-to-munder-difflin/): Connect Slack to Munder Difflin 0.5.2: create the Slack app, add the bot scopes, pick one of three connection modes, and get answers in the thread.+- [How to Connect Ollama to Claude Code](https://munderdiffl.in/blog/how-to-connect-ollama-to-claude-code/): Yes, Claude Code can run on Ollama. Run ollama launch claude or set three environment variables, give the model 64K of context, and see what stops working.+- [Codex CLI: What It Is and How to Install It on Mac, Windows and Linux](https://munderdiffl.in/blog/how-to-install-codex-cli/): Codex CLI is OpenAI's open source coding agent for the terminal. Install it on Mac, Windows or Linux, sign in, update it, check the version or remove it.+- [Why Does Claude Code Keep Asking for Permission?](https://munderdiffl.in/blog/why-does-claude-code-keep-asking-for-permission/): Claude Code keeps asking because Manual mode gates edits, shell commands and web requests. The fixes, safest first: modes, allow rules, sandboxing, then bypass.+- [Claude Code Planning Mode: How to Use Claude Plan Mode](https://munderdiffl.in/blog/how-to-use-claude-code-plan-mode/): Claude Code plan mode, checked 29 Sep 2026: the Shift+Tab shortcut, --permission-mode plan, defaultMode, what it blocks, the approval options and opusplan.+- [Claude Code MCP Servers: How to Add One, Scopes and .mcp.json](https://munderdiffl.in/blog/how-to-add-an-mcp-server-to-claude-code/): Add Claude Code MCP servers with claude mcp add: transports, scopes, --env and --header, .mcp.json, OAuth sign in and output limits. Checked 29 Sep 2026.+- [The Command Center: Kanban, Fleet and Budgets in One Place](https://munderdiffl.in/blog/command-center-guide/): A guide to Munder Difflin's Command Center in 0.5.2: the kanban with task dependencies, the ask me tab, triggers, memory, live token use against each agent's budget, and when to watch the board instead of the floor.+- [How to Add a GitHub Copilot CLI Agent to Your Floor](https://munderdiffl.in/blog/how-to-add-a-github-copilot-cli-agent/): A step-by-step v0.3.3 walkthrough: install GitHub Copilot CLI (or let the harness install it for you), hire a Copilot worker, pick a model, and know exactly what print mode can and can't do.+- [How to Hire From the Agent Gallery](https://munderdiffl.in/blog/how-to-hire-from-the-agent-gallery/): A practical guide to Munder Difflin's Agent Gallery: pick one of 80 ready made roles, download its manifest, import it with Add agent, review every field, customise identity, workspace, engine and briefing, then spawn it yourself.+- [How to Brief Your Orchestrator (So the Floor Actually Ships)](https://munderdiffl.in/blog/how-to-talk-to-your-orchestrator/): A practical guide to briefing Munder Difflin's GOD orchestrator: state the goal not the steps, set constraints and budgets, name the deliverable, and let Michael staff the floor. With bad-vs-good brief examples, mid-run steering, and Talk mode.+- [How to Use the Built-in Monaco IDE](https://munderdiffl.in/blog/how-to-use-the-built-in-monaco-ide/): A practical walkthrough of Munder Difflin v0.3.3's built-in Monaco IDE: the title-bar IDE button, the git CHANGES rail with side-by-side diffs vs HEAD, the file tree, tabs, Cmd/Ctrl+S save — and the agent review workflow it enables.+- [Run a Mixed-Engine Office: the Right CLI for Every Desk](https://munderdiffl.in/blog/run-a-mixed-engine-office/): A practical guide to mixing agent engines on one Munder Difflin floor: Claude Code as orchestrator, Codex for coding bursts, Copilot for dispatched tasks, and OpenCode/Crush/pi.dev for BYOK keys and local models.+- [Run Munder Difflin Locally on a Mac Mini](https://munderdiffl.in/blog/run-munder-difflin-on-a-mac-mini/): Run a whole Munder Difflin office offline on an Apple silicon Mac mini: size the model to your unified memory, install Ollama or LM Studio, and wire OpenCode, Crush, Qwen and Pi to it. Current as of Munder Difflin 0.5.2 and the M6 and M5 Pro Mac mini.+- [Run Munder Difflin on Open Source Models: Fully Local or Through a Provider](https://munderdiffl.in/blog/run-munder-difflin-on-open-models/): Munder Difflin can run your whole agent floor on open weight models like gpt-oss, Qwen3, DeepSeek, Llama, GLM and Kimi: fully local with Ollama, LM Studio or vLLM, or through a provider with your own key. The wiring for each engine, current as of 0.5.2.+- [Run Your AI Agent Hive from Slack: The Complete Setup](https://munderdiffl.in/blog/run-ai-agent-hive-from-slack-setup/): Step-by-step setup for driving Munder Difflin's AI agent hive from Slack: create the app, set scopes, paste the tunnel Request URL, and @mention your bot to start work — done-summaries post back in-thread.+- [How to Trigger Your AI Agent Hive from a GitHub Webhook](https://munderdiffl.in/blog/trigger-ai-agents-from-github-webhook/): Wire a GitHub webhook to Munder Difflin: a secret-gated local endpoint turns each repo event into a task for your GOD orchestrator — POST a message, get a token, poll the result. No server to host.+- [Munder Difflin v0.2.4 Feature Walkthrough](https://munderdiffl.in/blog/munder-difflin-v0-2-4-feature-walkthrough/): A comprehensive guide to every change in Munder Difflin v0.2.4 — how the Codex lifecycle-hook bridge achieves full hive parity, what the Schedules tab adds, why tunnelmole replaced localtunnel, and what else shipped.+- [Does the June 2026 Agent SDK Change Affect Munder Difflin?](https://munderdiffl.in/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/): Anthropic paused its June 15, 2026 Agent SDK billing change. Munder Difflin drives the native Claude Code CLI, so your hive runs on your plan as before.+- [MCP Tool Poisoning: How to Secure the Model Context Protocol](https://munderdiffl.in/blog/mcp-security-tool-poisoning/): How the Model Context Protocol gets attacked — tool poisoning, rug pulls, command injection — and the layered defenses that secure your MCP tools.+- [The Multi-Agent Cost Playbook: Five Levers to Cut AI Agent Spend](https://munderdiffl.in/blog/the-multi-agent-cost-playbook/): Caching, batching, model tiering, context discipline, local-first — the five levers that cut a multi-agent fleet's bill, and how they compound.+- [Running AI Agents Safely: Permission Modes and Sandboxing](https://munderdiffl.in/blog/agent-security-and-sandboxing/): How to run autonomous AI agents safely: permission modes and the bypassPermissions foot-gun, workspace isolation, and what to gate vs. allow.+- [Building Reliable AI Agents: Failure Is the Default](https://munderdiffl.in/blog/building-reliable-ai-agents/): Agents fail constantly: tools error, steps stall, output goes wrong. Reliability isn't avoiding failure — it's making each one contained and recoverable.+- [Claude Code Automation: Let Agents Build While You Sleep](https://munderdiffl.in/blog/claude-code-automation-while-you-sleep/): Claude Code automation that runs overnight: an autonomous Stop-hook loop, safe permission bypass, and guardrails so you wake up to progress, not chaos.+- [Context Engineering for AI Agents: Budgeting the Window](https://munderdiffl.in/blog/context-engineering-for-ai-agents/): Why context engineering beats prompt wording for agents — and the tactics (isolation, retrieval, externalized state, compaction) that keep the window lean.+- [How to Debug a Multi-Agent System](https://munderdiffl.in/blog/debugging-multi-agent-systems/): How to debug a multi-agent system: use the event log, per-agent terminals, message trails, and git history to find why a hive of AI agents went sideways.+- [Evaluating AI Agent Reliability: How to Measure What You Can Trust](https://munderdiffl.in/blog/evaluating-ai-agent-reliability/): How to measure whether an AI agent is reliable enough to trust: reliability thresholds, why benchmarks overstate, and evaluating on your own codebase.+- [Approving AI Agents: Human-in-the-Loop Without a Queue](https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/): How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.+- [Prompt Caching for AI Agents: Stop Paying for the Same Tokens Twice](https://munderdiffl.in/blog/prompt-caching-for-ai-agents/): An agent re-sends the same prompt, tools, and context every turn. Prompt caching makes you pay for that prefix once — here's how to design for it.+- [Security for AI Coding Agents: A Practical Guide](https://munderdiffl.in/blog/security-for-ai-coding-agents/): How to run AI coding agents safely: contain the blast radius, scope every task, treat agent input as untrusted, and gate the irreversible behind a human.+- [A Small-Team Playbook for AI Coding Agents](https://munderdiffl.in/blog/small-team-ai-agents-playbook/): How a small team adopts a hive of AI coding agents without chaos — the roles to assign, the cadence to run, and the guardrails that keep it safe.+- [The Lethal Trifecta for Coding Agents](https://munderdiffl.in/blog/the-lethal-trifecta-for-coding-agents/): A coding agent reads private code, ingests untrusted content, and runs commands — the lethal trifecta. How a poisoned dependency leaks your secrets.+- [Trigger Your AI Agent Hive from Slack](https://munderdiffl.in/blog/trigger-ai-agents-from-slack/): Trigger an AI agent hive from Slack: a local webhook verifies each message and drops it into your orchestrator's queue as a task — no server to host.+- [What's New in Agentic AI: A June 2026 Field Guide](https://munderdiffl.in/blog/whats-new-in-agentic-ai-june-2026/): A grounded June 2026 roundup of agentic AI: Claude Opus 4.8, the MCP and A2A protocol layer, usage-based agent billing, and agent governance.+- [Bringing Your MCP Servers and Skills Into a Hive of Agents](https://munderdiffl.in/blog/mcp-and-skills-in-a-hive/): How agents in a hive inherit your existing Claude Code MCP servers, skills, and tools — and how to scope which agent gets which capability.+- [How to Run Multiple Claude Code Agents (Without Losing Track)](https://munderdiffl.in/blog/how-to-run-multiple-claude-code-agents/): Run several Claude Code agents in parallel without the chaos: give each a role, let them coordinate, and stop alt-tabbing between terminal windows.+- [How to Manage Multiple Claude Code Sessions Like a Pro](https://munderdiffl.in/blog/manage-multiple-claude-code-sessions/): Tactics to manage multiple Claude Code sessions without losing track: naming, roles, context isolation, and when a harness beats juggling terminal tabs.+- [Set Up a Claude Code Multi-Agent Workflow in 10 Minutes](https://munderdiffl.in/blog/claude-code-multi-agent-setup-tutorial/): A from-zero Claude Code multi-agent setup: install Munder Difflin, onboard, spawn a few agents, and watch the GOD orchestrator route your first task.+- [Claude Code Subagents: How to Create and Use Them (vs Agent Teams)](https://munderdiffl.in/blog/claude-code-subagents-vs-multi-agent-harness/): Claude Code subagents in 2026: the .claude/agents file, built-in Explore and Plan, models, nesting, and when agent teams or a harness fit better.### concepts@@ -97,58 +150,6 @@- [What Are Claude Code Agents? And How to Use Many at Once](https://munderdiffl.in/blog/what-are-claude-code-agents/): Claude Code agents explained in plain English — what an agent actually is, how subagents differ, and the leap from one agent to a coordinated team.- [What Is an AI Agent Harness? Single vs Multi-Agent Harness Explained](https://munderdiffl.in/blog/what-is-a-multi-agent-harness/): An agent harness is the code around an AI model that runs its tool loop, memory and permissions. Single vs multi-agent harnesses, compared and dated.-### guides--- [How to Install and Use Munder Difflin: A Beginner's Guide](https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/): A plain language guide to Munder Difflin 0.5.2. What a coding agent is, which AI engine to pick based on what you already pay for, the one terminal command you'll ever run, and how to install and set it up on macOS, Windows or Linux.-- [Your First Hour With Munder Difflin: Onboarding to a Pro Office](https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/): Munder Difflin 0.5.2 from download to a working Pro office: install Claude Code, Codex or OpenCode, sign in, go Pro, then automate reviews, Slack and email.-- [How to Get the Most Out of Your Claude Code Max Plan](https://munderdiffl.in/blog/claude-code-max-plan-tips/): Stretch a Claude Code Max plan: model choice, context hygiene, prompt caching, subagents and unattended runs, checked against Anthropic's own pages.-- [How to Make the Most of Your Codex Max Plan](https://munderdiffl.in/blog/codex-max-plan-tips/): There is no Codex Max plan. What ChatGPT Pro 5x and 20x give you in Codex, how the weekly limit works, and how to stretch it.-- [Gemini 4 Argon Pricing: Half Price Until 30 November, Then the Real Bill Arrives](https://munderdiffl.in/blog/gemini-4-argon-pricing/): Gemini 4 Argon pricing, checked 13 Oct 2026: $2 input and $10 output per 1M tokens through 30 Nov, then $4 and $20. No free tier. Worked examples inside.-- [How to Use Gemini 4 Argon: Five Ways In, and Free Is Not One of Them](https://munderdiffl.in/blog/how-to-use-gemini-4-argon/): How to use Gemini 4 Argon on 13 Oct 2026: the Gemini app, AI Mode, AI Studio, the API with model ID gemini-4-argon, and Antigravity. What each route needs.-- [ASD-STE100 Prompting: Karpathy's 4 Tricks for Clearer AI Answers](https://munderdiffl.in/blog/asd-ste100-prompting/): Ask your AI to answer in ASD-STE100, the aerospace manual standard, about 80% of the way. Plus Karpathy's three other tricks: diagrams, HTML and video.-- [Codex Pricing: Every Plan, Usage Limit and API Rate (Oct 2026)](https://munderdiffl.in/blog/codex-pricing/): Codex pricing checked 5 Oct 2026: which ChatGPT plans include Codex, US prices from $0 to $500, how usage limits are counted, credits and API key rates.-- [How to Use Claude Code: Setup, First Task and the Commands That Matter](https://munderdiffl.in/blog/how-to-use-claude-code/): How to use Claude Code, checked 5 Oct 2026: install it, sign in, run a first task, then the commands, shortcuts and habits a beginner needs.-- [Build Your Personal AGI on Your Computer Using Claude Code and Munder Difflin](https://munderdiffl.in/blog/build-your-personal-agi-on-your-computer/): Turn Claude Code into a personal AGI on your own computer: agents with memory, schedules, Gmail, Calendar, voice and Slack, set up in Munder Difflin 0.5.2.-- [How to Install Gemini CLI, and Can It Use Gemini 4 Yet?](https://munderdiffl.in/blog/how-to-install-gemini-cli/): Install Gemini CLI 0.62.0 with one npm command, see which sign ins still work after 18 Jun 2026, and why Gemini 4 Argon isn't in it yet.-- [How to Connect Slack to Munder Difflin 0.5.2](https://munderdiffl.in/blog/connect-slack-to-munder-difflin/): Connect Slack to Munder Difflin 0.5.2: create the Slack app, add the bot scopes, pick one of three connection modes, and get answers in the thread.-- [How to Connect Ollama to Claude Code](https://munderdiffl.in/blog/how-to-connect-ollama-to-claude-code/): Yes, Claude Code can run on Ollama. Run ollama launch claude or set three environment variables, give the model 64K of context, and see what stops working.-- [Codex CLI: What It Is and How to Install It on Mac, Windows and Linux](https://munderdiffl.in/blog/how-to-install-codex-cli/): Codex CLI is OpenAI's open source coding agent for the terminal. Install it on Mac, Windows or Linux, sign in, update it, check the version or remove it.-- [Why Does Claude Code Keep Asking for Permission?](https://munderdiffl.in/blog/why-does-claude-code-keep-asking-for-permission/): Claude Code keeps asking because Manual mode gates edits, shell commands and web requests. The fixes, safest first: modes, allow rules, sandboxing, then bypass.-- [Claude Code Planning Mode: How to Use Claude Plan Mode](https://munderdiffl.in/blog/how-to-use-claude-code-plan-mode/): Claude Code plan mode, checked 29 Sep 2026: the Shift+Tab shortcut, --permission-mode plan, defaultMode, what it blocks, the approval options and opusplan.-- [Claude Code MCP Servers: How to Add One, Scopes and .mcp.json](https://munderdiffl.in/blog/how-to-add-an-mcp-server-to-claude-code/): Add Claude Code MCP servers with claude mcp add: transports, scopes, --env and --header, .mcp.json, OAuth sign in and output limits. Checked 29 Sep 2026.-- [The Command Center: Kanban, Fleet and Budgets in One Place](https://munderdiffl.in/blog/command-center-guide/): A guide to Munder Difflin's Command Center in 0.5.2: the kanban with task dependencies, the ask me tab, triggers, memory, live token use against each agent's budget, and when to watch the board instead of the floor.-- [How to Add a GitHub Copilot CLI Agent to Your Floor](https://munderdiffl.in/blog/how-to-add-a-github-copilot-cli-agent/): A step-by-step v0.3.3 walkthrough: install GitHub Copilot CLI (or let the harness install it for you), hire a Copilot worker, pick a model, and know exactly what print mode can and can't do.-- [How to Hire From the Agent Gallery](https://munderdiffl.in/blog/how-to-hire-from-the-agent-gallery/): A practical guide to Munder Difflin's Agent Gallery: pick one of 80 ready made roles, download its manifest, import it with Add agent, review every field, customise identity, workspace, engine and briefing, then spawn it yourself.-- [How to Brief Your Orchestrator (So the Floor Actually Ships)](https://munderdiffl.in/blog/how-to-talk-to-your-orchestrator/): A practical guide to briefing Munder Difflin's GOD orchestrator: state the goal not the steps, set constraints and budgets, name the deliverable, and let Michael staff the floor. With bad-vs-good brief examples, mid-run steering, and Talk mode.-- [How to Use the Built-in Monaco IDE](https://munderdiffl.in/blog/how-to-use-the-built-in-monaco-ide/): A practical walkthrough of Munder Difflin v0.3.3's built-in Monaco IDE: the title-bar IDE button, the git CHANGES rail with side-by-side diffs vs HEAD, the file tree, tabs, Cmd/Ctrl+S save — and the agent review workflow it enables.-- [Run a Mixed-Engine Office: the Right CLI for Every Desk](https://munderdiffl.in/blog/run-a-mixed-engine-office/): A practical guide to mixing agent engines on one Munder Difflin floor: Claude Code as orchestrator, Codex for coding bursts, Copilot for dispatched tasks, and OpenCode/Crush/pi.dev for BYOK keys and local models.-- [Run Munder Difflin Locally on a Mac Mini](https://munderdiffl.in/blog/run-munder-difflin-on-a-mac-mini/): Run a whole Munder Difflin office offline on an Apple silicon Mac mini: size the model to your unified memory, install Ollama or LM Studio, and wire OpenCode, Crush, Qwen and Pi to it. Current as of Munder Difflin 0.5.2 and the M6 and M5 Pro Mac mini.-- [Run Munder Difflin on Open Source Models: Fully Local or Through a Provider](https://munderdiffl.in/blog/run-munder-difflin-on-open-models/): Munder Difflin can run your whole agent floor on open weight models like gpt-oss, Qwen3, DeepSeek, Llama, GLM and Kimi: fully local with Ollama, LM Studio or vLLM, or through a provider with your own key. The wiring for each engine, current as of 0.5.2.-- [Run Your AI Agent Hive from Slack: The Complete Setup](https://munderdiffl.in/blog/run-ai-agent-hive-from-slack-setup/): Step-by-step setup for driving Munder Difflin's AI agent hive from Slack: create the app, set scopes, paste the tunnel Request URL, and @mention your bot to start work — done-summaries post back in-thread.-- [How to Trigger Your AI Agent Hive from a GitHub Webhook](https://munderdiffl.in/blog/trigger-ai-agents-from-github-webhook/): Wire a GitHub webhook to Munder Difflin: a secret-gated local endpoint turns each repo event into a task for your GOD orchestrator — POST a message, get a token, poll the result. No server to host.-- [Munder Difflin v0.2.4 Feature Walkthrough](https://munderdiffl.in/blog/munder-difflin-v0-2-4-feature-walkthrough/): A comprehensive guide to every change in Munder Difflin v0.2.4 — how the Codex lifecycle-hook bridge achieves full hive parity, what the Schedules tab adds, why tunnelmole replaced localtunnel, and what else shipped.-- [Does the June 2026 Agent SDK Change Affect Munder Difflin?](https://munderdiffl.in/blog/does-the-june-2026-agent-sdk-change-affect-munder-difflin/): Anthropic paused its June 15, 2026 Agent SDK billing change. Munder Difflin drives the native Claude Code CLI, so your hive runs on your plan as before.-- [MCP Tool Poisoning: How to Secure the Model Context Protocol](https://munderdiffl.in/blog/mcp-security-tool-poisoning/): How the Model Context Protocol gets attacked — tool poisoning, rug pulls, command injection — and the layered defenses that secure your MCP tools.-- [The Multi-Agent Cost Playbook: Five Levers to Cut AI Agent Spend](https://munderdiffl.in/blog/the-multi-agent-cost-playbook/): Caching, batching, model tiering, context discipline, local-first — the five levers that cut a multi-agent fleet's bill, and how they compound.-- [Running AI Agents Safely: Permission Modes and Sandboxing](https://munderdiffl.in/blog/agent-security-and-sandboxing/): How to run autonomous AI agents safely: permission modes and the bypassPermissions foot-gun, workspace isolation, and what to gate vs. allow.-- [Building Reliable AI Agents: Failure Is the Default](https://munderdiffl.in/blog/building-reliable-ai-agents/): Agents fail constantly: tools error, steps stall, output goes wrong. Reliability isn't avoiding failure — it's making each one contained and recoverable.-- [Claude Code Automation: Let Agents Build While You Sleep](https://munderdiffl.in/blog/claude-code-automation-while-you-sleep/): Claude Code automation that runs overnight: an autonomous Stop-hook loop, safe permission bypass, and guardrails so you wake up to progress, not chaos.-- [Context Engineering for AI Agents: Budgeting the Window](https://munderdiffl.in/blog/context-engineering-for-ai-agents/): Why context engineering beats prompt wording for agents — and the tactics (isolation, retrieval, externalized state, compaction) that keep the window lean.-- [How to Debug a Multi-Agent System](https://munderdiffl.in/blog/debugging-multi-agent-systems/): How to debug a multi-agent system: use the event log, per-agent terminals, message trails, and git history to find why a hive of AI agents went sideways.-- [Evaluating AI Agent Reliability: How to Measure What You Can Trust](https://munderdiffl.in/blog/evaluating-ai-agent-reliability/): How to measure whether an AI agent is reliable enough to trust: reliability thresholds, why benchmarks overstate, and evaluating on your own codebase.-- [Approving AI Agents: Human-in-the-Loop Without a Queue](https://munderdiffl.in/blog/human-in-the-loop-approving-ai-agents/): How to approve what AI agents do without a parallel approval queue: native permission prompts, routing 'to: human' to the orchestrator, remote approval.-- [Prompt Caching for AI Agents: Stop Paying for the Same Tokens Twice](https://munderdiffl.in/blog/prompt-caching-for-ai-agents/): An agent re-sends the same prompt, tools, and context every turn. Prompt caching makes you pay for that prefix once — here's how to design for it.-- [Security for AI Coding Agents: A Practical Guide](https://munderdiffl.in/blog/security-for-ai-coding-agents/): How to run AI coding agents safely: contain the blast radius, scope every task, treat agent input as untrusted, and gate the irreversible behind a human.-- [A Small-Team Playbook for AI Coding Agents](https://munderdiffl.in/blog/small-team-ai-agents-playbook/): How a small team adopts a hive of AI coding agents without chaos — the roles to assign, the cadence to run, and the guardrails that keep it safe.-- [The Lethal Trifecta for Coding Agents](https://munderdiffl.in/blog/the-lethal-trifecta-for-coding-agents/): A coding agent reads private code, ingests untrusted content, and runs commands — the lethal trifecta. How a poisoned dependency leaks your secrets.-- [Trigger Your AI Agent Hive from Slack](https://munderdiffl.in/blog/trigger-ai-agents-from-slack/): Trigger an AI agent hive from Slack: a local webhook verifies each message and drops it into your orchestrator's queue as a task — no server to host.-- [What's New in Agentic AI: A June 2026 Field Guide](https://munderdiffl.in/blog/whats-new-in-agentic-ai-june-2026/): A grounded June 2026 roundup of agentic AI: Claude Opus 4.8, the MCP and A2A protocol layer, usage-based agent billing, and agent governance.-- [Bringing Your MCP Servers and Skills Into a Hive of Agents](https://munderdiffl.in/blog/mcp-and-skills-in-a-hive/): How agents in a hive inherit your existing Claude Code MCP servers, skills, and tools — and how to scope which agent gets which capability.-- [How to Run Multiple Claude Code Agents (Without Losing Track)](https://munderdiffl.in/blog/how-to-run-multiple-claude-code-agents/): Run several Claude Code agents in parallel without the chaos: give each a role, let them coordinate, and stop alt-tabbing between terminal windows.-- [How to Manage Multiple Claude Code Sessions Like a Pro](https://munderdiffl.in/blog/manage-multiple-claude-code-sessions/): Tactics to manage multiple Claude Code sessions without losing track: naming, roles, context isolation, and when a harness beats juggling terminal tabs.-- [Set Up a Claude Code Multi-Agent Workflow in 10 Minutes](https://munderdiffl.in/blog/claude-code-multi-agent-setup-tutorial/): A from-zero Claude Code multi-agent setup: install Munder Difflin, onboard, spawn a few agents, and watch the GOD orchestrator route your first task.-- [Claude Code Subagents: How to Create and Use Them (vs Agent Teams)](https://munderdiffl.in/blog/claude-code-subagents-vs-multi-agent-harness/): Claude Code subagents in 2026: the .claude/agents file, built-in Explore and Plan, models, nesting, and when agent teams or a harness fit better.-### comparisons- [Grok Bot Alternatives: 9 Picks, Hosted or on Your Computer](https://munderdiffl.in/blog/grok-bot-alternatives/): Grok Bot runs every Bot on one shared cloud computer in the US. Nine alternatives checked on 30 Sep 2026, from ChatGPT dots to open source agents you run.diff --git a/docs/sitemap.xml b/docs/sitemap.xmlindex 4a83bb24..9ce879b5 100644--- a/docs/sitemap.xml+++ b/docs/sitemap.xml@@ -10,7 +10,7 @@<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"><url><loc>https://munderdiffl.in/</loc>- <lastmod>2026-10-13</lastmod>+ <lastmod>2026-10-14</lastmod><changefreq>weekly</changefreq><priority>1.0</priority></url>@@ -21,6 +21,12 @@<url><loc>https://munderdiffl.in/blog/</loc><changefreq>daily</changefreq><priority>0.9</priority></url><url><loc>https://munderdiffl.in/blog/topics/</loc><changefreq>weekly</changefreq><priority>0.6</priority></url><url><loc>https://munderdiffl.in/blog/about/</loc><changefreq>monthly</changefreq><priority>0.4</priority></url>+ <url>+ <loc>https://munderdiffl.in/blog/ai-agent-use-cases-our-office/</loc>+ <lastmod>2026-10-14</lastmod>+ <changefreq>monthly</changefreq>+ <priority>0.8</priority>+ </url><url><loc>https://munderdiffl.in/blog/agent-tools-today-2026-10-13/</loc><lastmod>2026-10-13</lastmod>@@ -701,7 +707,7 @@</url><url><loc>https://munderdiffl.in/blog/your-first-hour-with-munder-difflin/</loc>- <lastmod>2026-09-14</lastmod>+ <lastmod>2026-10-14</lastmod><changefreq>monthly</changefreq><priority>0.8</priority></url>@@ -839,7 +845,7 @@</url><url><loc>https://munderdiffl.in/blog/how-to-install-and-use-munder-difflin/</loc>- <lastmod>2026-09-10</lastmod>+ <lastmod>2026-10-14</lastmod><changefreq>monthly</changefreq><priority>0.8</priority></url>@@ -1293,8 +1299,8 @@<changefreq>monthly</changefreq><priority>0.8</priority></url>- <url><loc>https://munderdiffl.in/blog/topics/concepts/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url><url><loc>https://munderdiffl.in/blog/topics/guides/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url>+ <url><loc>https://munderdiffl.in/blog/topics/concepts/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url><url><loc>https://munderdiffl.in/blog/topics/comparisons/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url><url><loc>https://munderdiffl.in/blog/topics/internals/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url><url><loc>https://munderdiffl.in/blog/topics/story/</loc><changefreq>weekly</changefreq><priority>0.5</priority></url>