Skip to content

Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.

fix(forge-nim): render a finished reconstruction and say when no NIM is wired

nirholas555dafb9a218
diff --git a/CHANGELOG.md b/CHANGELOG.md
index c32a15355..9f3428798 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
## 2026-08-17
- **Docs · Cars in /play: traffic and the drivable fleet** (`/docs/play-vehicles`): The Trench Car is the default car in every coin world: ambient traffic drives it and it is parked for you to take the wheel. How to drive, how a car is assembled from spec, physics and mesh, how the model is staged, and how to add another.
+- **A busy Forge engine now says it is busy and counts you back in, instead of telling you to go set an environment variable**: The /forge page was driven end to end in a real browser. When every free 3D engine was momentarily at capacity, or a worker was still warming up, the server said so clearly and told the page to try again in about half a minute. The page threw that away. It treated any temporary outage as proof that the whole generator had never been set up, and showed a panel headed Generator not configured that asked you to set NVIDIA_API_KEY or HF_TOKEN on the deployment. For anyone who is not running the servers, which is everyone using the site, that is an instruction you cannot act on, about a problem you do not have, and the panel had no Try again button at all, so a wait of thirty seconds became a dead end that looked like the product was broken. A temporary outage now lands in the wait state the site already had for this: a heading that says Temporarily unavailable, the countdown ticking down to when the engines free up, and a Retry button that switches itself on the moment the wait is over. The genuine case, a deployment with no engine wired at all, still gets its original panel, because for the operator that message is the correct one. The engine picker also had a key underneath it explaining that a red dot means an engine is unavailable, but no engine could ever show that dot: only the amber busy dot was ever drawn, so a down engine was dimmed with no indication why, and the key described something that did not exist. Down engines now carry the red dot the key promises, at full strength so it stays distinct from the amber one. Separately, roughly one in ten preview thumbnails failed to load while developing locally, which made the gallery look like a broken image host when nothing was actually wrong with it. (`/forge`) `[fix, improvement]`
- **A failed X seed no longer costs you six hours, and can no longer wipe a good one**: Seeding an agent from your X posts is allowed once every six hours per agent, and until now that window was spent the moment we started reading, whatever happened next. If X was throttling us, if X had stopped accepting the connection, or if your sign-in could not be renewed, you got a bare failure and a six-hour lockout for something that was never yours to fix. Every one of those now hands the window straight back and says what actually happened: reconnect X, or wait the number of minutes X itself told us to wait. Worse was the run that read your account fine and found nothing usable in it, which is what a timeline of only replies and reposts looks like. That run deleted the batch of facts your agent already had and put nothing in its place. It now stops before the deletion, keeps the memories you had, gives the window back, and tells you what to post or add to your profile so the next seed has something to work with. Nothing else about the lane changes: nothing is read until you consent on a screen that lists what is read and stored, and revoking still deletes every memory the seeding produced. (`/settings`) `[fix]`
- **A shared ASL letter link now opens on that letter, not on A**: The manual alphabet at /asl-alphabet was audited end to end in a real browser, and six repairs came out of it. Sharing a letter was the worst of them: a /asl-alphabet?letter=W link built the whole page around the letter A and only corrected itself once the 3D avatar had finished downloading, so anyone on a slow connection read A's description, saw A's key lit and A's look-alike note for the entire wait, on a link that promised W. The page now reads the link before it builds anything, so the right letter is on the card from the first paint and the avatar simply catches up. The Spell anything box had the opposite problem: it reported an empty or unspellable word only in the small caption pinned to the avatar, which on a phone sits a full screen above the box you just typed in, so pressing Spell it with a word full of punctuation looked like a dead button. It now answers in its own line, right under the input, and while the hand is spelling it lays the word out letter by letter and lights each one as the hand reaches it, so a long word can be followed without scrolling back up to the A-to-Z grid. That caption itself was white text on a scrim too faint to read over a pale avatar or in light mode, and is now a proper legible chip. Opening the avatar picker no longer lets a keystroke through to sign a letter behind the dialog. Switching avatars used to leave 'Loading avatar…' on screen forever once the new one had arrived, which also meant screen readers were never told the load had finished. And the Spell panel is no longer stretched to the height of the drill beside it, which left a large empty box on desktop. (`/asl-alphabet`) `[fix, improvement]`
- **Avatar Studio now shows your whole avatar, and its Save button fits on a phone**: The studio camera was framing the base body from the shoulders to the knees: the face you were sculpting and the shoes you were picking were both outside the picture, and the only way to see them was to orbit the camera yourself every time the page loaded. The framing is now solved from the lens rather than guessed at, so a full-body shot really does hold the avatar head to toe, arms included, on any window shape, and the half-body and headshot views stay as tight as their names promise. On a 320px phone the top bar had been running past the edge of the screen and taking the Save avatar button with it, so a look built on a phone could not be saved at all; the bar now fits, with the Back and title labels standing down to make room. The tab strip on that same narrow layout had been hiding its labels without putting anything in their place, leaving six blank buttons, and now shows an icon per tab. The Parametric body badge that read 120+ sliders was white text on a white pill and invisible everywhere. A failed save now says the connection could not be reached and that your look is still there, instead of repeating the browser phrase Failed to fetch, and every accessory search box, every tab and the Back link now carry names a screen reader can read. Opening avatar-studio in local development also serves the same page production does, which it previously did not. (`/avatar-studio`) `[fix, improvement]`
@@ -30,6 +31,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
- **Scene Composer works on a phone, starts you off with real models, and answers the keyboard**: Scene Composer was unusable on anything narrower than a laptop. Below 900px the scene panel was switched off, which pushed the 3D canvas into a zero-width column, so a tablet visitor got no viewport at all and no way back to the hierarchy, inspector or bone-attach controls; at phone widths the toolbar forced the whole editor wider than the screen and shoved the right-hand panel out of view. The panel is now a drawer behind a toolbar button, the toolbar scrolls instead of stretching the layout, and the canvas keeps real dimensions at 320, 768 and 1440 pixels. The Recent Creations strip also no longer greets a first-time visitor with an empty box: until you have forged something it shows the live community showcase, so there are real models to drop into the scene straight away. The avatar picker was asking the directory a question it does not answer and got back a feed of on-chain agents with no 3D model attached, so most slots were silently dropped and the rest never showed their thumbnails; it now asks for public 3D avatars and renders them. The start panel finally honours what it always promised: paste an avatar ID, not just a URL, and a model that cannot be loaded leaves a written explanation instead of a toast that vanishes before you can read it. Both the creations strip and the avatar picker now offer a Try again that works when a request fails. On the keyboard side: the page has a real heading, the camera-view menu and every creation and avatar tile are proper buttons, the scene rows expose their hide and remove actions on focus, the prompt and avatar fields have labels, dialogs trap focus and hand it back on Escape, and there is a visible focus ring throughout. Removing an object now frees its GPU memory once the step drops out of undo range, undo puts a bone-attached item back on its bone, and saving an outfit refuses a locally dropped file rather than storing a link that breaks on reload. (`/compose`) `[fix, improvement]`
- **Self-hosted NIM says up front that no endpoint is wired, and a finished reconstruction actually appears on screen**: The /forge-nim page drives a TRELLIS NIM you host yourself, and it was driven end to end in a real browser. It already asked the API on load whether a NIM was configured, then threw the answer away: there was nowhere on the page to show it. So a first-time visitor dropped a photo, waited, and got a failure at the end of the trip, and the failure read as the raw code nim_unconfigured because the page was reading the machine code instead of the sentence next to it. The page now carries a status pill that says whether a NIM is wired and reachable, and when none is, the stage explains that this page needs a NIM you host, offers a button that opens the endpoint field, and links to the hosted forge and the NVIDIA model docs in the meantime; the Reconstruct button stays switched off rather than promising a trip that cannot arrive. Typing your own endpoint now re-probes that box and reports what it found, and an address the server refuses comes back with the reason instead of a silent failure later. The bigger fix was invisible until a reconstruction finished: the 3D viewer had collapsed to zero height, so the model was never even loaded and a completed job showed an empty stage under its own result bar. The viewer now fills the stage. The button also never showed its loading spinner, because translating the page had quietly replaced the label it was written into. The page carries a real heading, the loading and error states announce themselves to screen readers, every control has a visible keyboard focus ring, and it holds up at phone width. (`/forge-nim`) `[fix, improvement]`
- **Selfie to Avatar tells you your library is full before it spends ninety seconds building**: Five fixes to Selfie to Avatar. If your avatar library was already at the limit for your plan, nothing noticed until the very end: you waited out the full reconstruction, and then the page said the 3D engine had hit a snag and told you to try again. It was not an engine problem, trying again could never work, and every attempt spent another ninety seconds of generation to land in the same place. Your library is now checked the moment you press the button, so a full one is refused immediately with the actual reason and what to do about it, and a build that fills the last slot while it is running ends with that same plain explanation instead of the engine message. The page also stopped losing its own error messages: the Build button floats above the page, so tapping it from higher up left the explanation somewhere below the fold while the button quietly reset itself, which read as nothing having happened. Errors now scroll into view. On the finished-avatar screen, Learn how to embed pointed at a page that did not exist and answered 404; it now opens the embedding guide. And on a narrow phone the Photoreal and Stylized choices were squeezed into two columns barely wide enough for one word per line, so they now stack and can be read. (`/create/selfie`) `[fix, improvement]`
+- **Skill tags now actually filter, a failed skills load says so instead of claiming the catalog is empty, and a shared skills link brings its filters with it**: The /skills path was driven end to end in a real browser. It forwards into the Marketplace Skills tab, and it was forwarding a search that the tab then threw away: a link like /skills?q=whale landed on the full unfiltered list with an empty search box. The Skills tab now reads its search, category, tag, and sort straight off the address bar, and writes them back as you change them, so the view you are looking at is the view you can bookmark, share, reload, and step back through. The tag chips on a skill were the clearest casualty of that gap: clicking one navigated to a list that looked exactly the same as before, because nothing was filtering by tag at all. Tags now filter for real in the catalog itself, the tab shows which tag it is filtering by, and one click clears it. When the catalog cannot be reached, the grid used to show the words No skills found with an invitation to publish one, which is the same screen an empty catalog would give you and offers no way to try again: the error state and its Retry button were being overwritten a fraction of a second after they rendered. A failed load now says it could not load skills and hands you the retry. The count under the heading was also being reverted to its placeholder by the translation pass a moment after it was written, so it never showed how many skills you were looking at; it does now, and it names your active category or tag. The empty state names the filters that emptied the list instead of guessing. For anyone arriving without JavaScript, including crawlers, /skills was a completely blank page with no link out of it; it now carries a real heading, an explanation, and a link into the Marketplace. The search box has a proper label, the free and paid chips announce which one is selected, and the page holds up from 320px to 1440px. (`/marketplace?tab=skills`) `[fix, improvement]`
- **Talking Avatar Video stops handing the renderer a 3D file where a photo belongs, and says up front when the GPU lane is down**: The /create/video page turns one of your avatars plus an audio clip into a lip-synced video, and it was driven end to end in a real browser: every state, every button, every width. The worst finding was invisible from the page. The reference frame the GPU worker renders from was being taken from whatever the browser had handy, which for an avatar with no thumbnail meant its 3D model file. The worker saved that as the reference photo and then failed on it minutes later, after the wait had already been spent. The page now sends only the avatar, and the server resolves a real image for it: the stored thumbnail, or a fresh portrait render when there is none. Second, the GPU renderer is a separate service, and when it is not running the only signal was a failure after you had already picked a file and uploaded it. The page now asks first and, if the lane is down, says so at the top, keeps Generate switched off, and points at the Avatar Studio and prompt-to-avatar in the meantime. The rest is the ordinary polish an audit turns up: a failed avatar list used to look identical to owning no avatars, so it sent people off to create one they already had, and now it says what broke and offers a retry; having no avatars yet gets a real empty state with a link to make one; the Generate button always explains what it is still waiting for and, once audio is in, how many clips of video that audio is worth; the clip you picked can be played back with its length and size before you spend a generation on it; a render job that goes unreachable now says so after thirty seconds instead of spinning for twenty minutes; and after a spent free trial the button under the error no longer stayed stuck on Upgrade plan for every later error. Avatar thumbnails are real buttons with visible focus rings, the two form fields have proper labels, and the second animated avatar that the site companion used to drop onto this page no longer loads over the one you are working on. (`/create/video`) `[fix, improvement]`
- **Text to 3D gets its footer back, a retry button that knows when to open, and an honest name for the model that refined your prompt**: The /forge-spark page was driven end to end in a real browser. The closing note under the workspace was written as a page footer, and the site's old corner-credit styling for the 3D viewer pages had quietly claimed it: it was pulled out of the layout into a floating monospace strip that sat on top of the viewer on a desktop, and on a 320px phone it landed across the composer with its link to the full studio pushed off the edge of the screen entirely. That note is now ordinary page text, centered under the workspace where it belongs, and the real three.ws footer, with its full set of links, is on the page for the first time. When a generation fails there is now a Try again button inside the error itself, so recovery does not mean scrolling back to the top; when the generator answers that it is busy, the button counts down and opens exactly when a retry can actually get through. A stage that fails is marked in red instead of silently dropping back to grey, the wait now shows real elapsed seconds so a cold reconstruction lane no longer reads as a hang, and the first step is renamed to whatever model actually answered, because the prompt refiner is a chain of engines and Nemotron is only the first one it tries. (`/forge-spark`) `[fix, improvement]`
- **The avatar SDK bundle we ship is now built from the code we deploy**: The <agent-3d> runtime that sites load from three.ws is a single bundled file, and the production build never rebuilt it. It was copied from whatever the build machine happened to have on disk, which could be days older than the code being deployed, so a fix to the avatar element could ship everywhere else on the site and quietly not reach the SDK. The deploy now rebuilds that bundle from the same commit as the rest of the release, every time. Nothing to change on your side: existing embeds keep working and simply pick up the current runtime. (`/avatar-sdk`) `[fix, sdk, infra]`
diff --git a/data/changelog.json b/data/changelog.json
index 52c1c12ff..c860368d1 100644
--- a/data/changelog.json
+++ b/data/changelog.json
@@ -1,6 +1,46 @@
{
"$comment": "Curated changelog entries for three.ws, the editorial layer of the public changelog. New PAGE launches are tracked automatically via the `added` field in data/pages.json; this file is for everything else holders care about: improvements, fixes, SDK releases, trading/payments capabilities, security work. scripts/build-page-index.mjs merges both sources into CHANGELOG.md, public/changelog.json, and public/changelog.xml. Newest first. Allowed tags: feature, improvement, fix, sdk, infra, docs, security. `link` is optional and must be a live three.ws path.",
"entries": [
+ {
+ "date": "2026-08-17",
+ "title": "A busy Forge engine now says it is busy and counts you back in, instead of telling you to go set an environment variable",
+ "summary": "The /forge page was driven end to end in a real browser. When every free 3D engine was momentarily at capacity, or a worker was still warming up, the server said so clearly and told the page to try again in about half a minute. The page threw that away. It treated any temporary outage as proof that the whole generator had never been set up, and showed a panel headed Generator not configured that asked you to set NVIDIA_API_KEY or HF_TOKEN on the deployment. For anyone who is not running the servers, which is everyone using the site, that is an instruction you cannot act on, about a problem you do not have, and the panel had no Try again button at all, so a wait of thirty seconds became a dead end that looked like the product was broken. A temporary outage now lands in the wait state the site already had for this: a heading that says Temporarily unavailable, the countdown ticking down to when the engines free up, and a Retry button that switches itself on the moment the wait is over. The genuine case, a deployment with no engine wired at all, still gets its original panel, because for the operator that message is the correct one. The engine picker also had a key underneath it explaining that a red dot means an engine is unavailable, but no engine could ever show that dot: only the amber busy dot was ever drawn, so a down engine was dimmed with no indication why, and the key described something that did not exist. Down engines now carry the red dot the key promises, at full strength so it stays distinct from the amber one. Separately, roughly one in ten preview thumbnails failed to load while developing locally, which made the gallery look like a broken image host when nothing was actually wrong with it.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/forge"
+ },
+ {
+ "date": "2026-08-17",
+ "title": "Skill tags now actually filter, a failed skills load says so instead of claiming the catalog is empty, and a shared skills link brings its filters with it",
+ "summary": "The /skills path was driven end to end in a real browser. It forwards into the Marketplace Skills tab, and it was forwarding a search that the tab then threw away: a link like /skills?q=whale landed on the full unfiltered list with an empty search box. The Skills tab now reads its search, category, tag, and sort straight off the address bar, and writes them back as you change them, so the view you are looking at is the view you can bookmark, share, reload, and step back through. The tag chips on a skill were the clearest casualty of that gap: clicking one navigated to a list that looked exactly the same as before, because nothing was filtering by tag at all. Tags now filter for real in the catalog itself, the tab shows which tag it is filtering by, and one click clears it. When the catalog cannot be reached, the grid used to show the words No skills found with an invitation to publish one, which is the same screen an empty catalog would give you and offers no way to try again: the error state and its Retry button were being overwritten a fraction of a second after they rendered. A failed load now says it could not load skills and hands you the retry. The count under the heading was also being reverted to its placeholder by the translation pass a moment after it was written, so it never showed how many skills you were looking at; it does now, and it names your active category or tag. The empty state names the filters that emptied the list instead of guessing. For anyone arriving without JavaScript, including crawlers, /skills was a completely blank page with no link out of it; it now carries a real heading, an explanation, and a link into the Marketplace. The search box has a proper label, the free and paid chips announce which one is selected, and the page holds up from 320px to 1440px.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/marketplace?tab=skills"
+ },
+ {
+ "date": "2026-08-17",
+ "title": "Self-hosted NIM says up front that no endpoint is wired, and a finished reconstruction actually appears on screen",
+ "summary": "The /forge-nim page drives a TRELLIS NIM you host yourself, and it was driven end to end in a real browser. It already asked the API on load whether a NIM was configured, then threw the answer away: there was nowhere on the page to show it. So a first-time visitor dropped a photo, waited, and got a failure at the end of the trip, and the failure read as the raw code nim_unconfigured because the page was reading the machine code instead of the sentence next to it. The page now carries a status pill that says whether a NIM is wired and reachable, and when none is, the stage explains that this page needs a NIM you host, offers a button that opens the endpoint field, and links to the hosted forge and the NVIDIA model docs in the meantime; the Reconstruct button stays switched off rather than promising a trip that cannot arrive. Typing your own endpoint now re-probes that box and reports what it found, and an address the server refuses comes back with the reason instead of a silent failure later. The bigger fix was invisible until a reconstruction finished: the 3D viewer had collapsed to zero height, so the model was never even loaded and a completed job showed an empty stage under its own result bar. The viewer now fills the stage. The button also never showed its loading spinner, because translating the page had quietly replaced the label it was written into. The page carries a real heading, the loading and error states announce themselves to screen readers, every control has a visible keyboard focus ring, and it holds up at phone width.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/forge-nim"
+ },
+ {
+ "date": "2026-08-17",
+ "title": "Text to 3D gets its footer back, a retry button that knows when to open, and an honest name for the model that refined your prompt",
+ "summary": "The /forge-spark page was driven end to end in a real browser. The closing note under the workspace was written as a page footer, and the site's old corner-credit styling for the 3D viewer pages had quietly claimed it: it was pulled out of the layout into a floating monospace strip that sat on top of the viewer on a desktop, and on a 320px phone it landed across the composer with its link to the full studio pushed off the edge of the screen entirely. That note is now ordinary page text, centered under the workspace where it belongs, and the real three.ws footer, with its full set of links, is on the page for the first time. When a generation fails there is now a Try again button inside the error itself, so recovery does not mean scrolling back to the top; when the generator answers that it is busy, the button counts down and opens exactly when a retry can actually get through. A stage that fails is marked in red instead of silently dropping back to grey, the wait now shows real elapsed seconds so a cold reconstruction lane no longer reads as a hang, and the first step is renamed to whatever model actually answered, because the prompt refiner is a chain of engines and Nemotron is only the first one it tries.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/forge-spark"
+ },
{
"date": "2026-08-17",
"title": "Talking Avatar Video stops handing the renderer a 3D file where a photo belongs, and says up front when the GPU lane is down",
diff --git a/pages/forge-nim.html b/pages/forge-nim.html
index 1ff0547dd..65e4a7361 100644
--- a/pages/forge-nim.html
+++ b/pages/forge-nim.html
@@ -475,7 +475,13 @@
min-height: 420px;
background: radial-gradient(120% 120% at 50% 18%, #141b2e 0%, #0b0e18 55%, #05060c 100%);
}
+ /* Absolute fill, not height:100%. `.stage-view` is a flex item with an auto
+ specified height, so a percentage height on this child resolves to auto:
+ the viewer collapsed to 0px, and model-viewer's own visibility observer
+ then never loaded the GLB, so a finished reconstruction rendered nothing. */
model-viewer {
+ position: absolute;
+ inset: 0;
width: 100%;
height: 100%;
--poster-color: transparent;
@@ -724,8 +730,15 @@
</div>
</div>
- <button class="btn" id="generate" type="submit" data-i18n-html="forge_nim.reconstruct_3d">
- <span id="generate-label">Reconstruct 3D</span>
+ <!-- Both labels live in the markup and toggle with `hidden`. The button
+ itself carries no data-i18n-html: that key held the label's own
+ markup, so translating it replaced this subtree and detached the
+ node the script had captured, leaving the label frozen and the
+ in-flight spinner invisible. -->
+ <button class="btn" id="generate" type="submit">
+ <span class="spinner" id="generate-spinner" hidden aria-hidden="true"></span>
+ <span id="label-image" data-i18n="forge_nim.label_reconstruct">Reconstruct 3D</span>
+ <span id="label-text" hidden data-i18n="forge_nim.label_generate">Generate 3D</span>
</button>
<details class="advanced">
@@ -809,7 +822,7 @@
</div>
<!-- Loading -->
- <div class="stage-state" id="state-loading">
+ <div class="stage-state" id="state-loading" role="status" aria-live="polite">
<div class="state-inner">
<div class="loader-ring"></div>
<h3 id="loading-title" data-i18n="forge_nim.reconstructing">Reconstructing…</h3>
@@ -820,7 +833,7 @@
</div>
<!-- Error -->
- <div class="stage-state" id="state-error">
+ <div class="stage-state" id="state-error" role="alert">
<div class="state-inner">
<svg class="state-icon" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">
<path d="M10.29 3.86 1.82 18a2 2 0 0 0 1.71 3h16.94a2 2 0 0 0 1.71-3L13.71 3.86a2 2 0 0 0-3.42 0Z" />
@@ -870,7 +883,9 @@
prompt: $('prompt'),
baseUrl: $('base-url'),
generate: $('generate'),
- generateLabel: $('generate-label'),
+ labelImage: $('label-image'),
+ labelText: $('label-text'),
+ spinner: $('generate-spinner'),
viewer: $('viewer'),
stateIdle: $('state-idle'),
stateLoading: $('state-loading'),
@@ -1002,9 +1017,17 @@
els.tabText.setAttribute('aria-selected', String(!isImage));
els.paneImage.classList.toggle('active', isImage);
els.paneText.classList.toggle('active', !isImage);
- els.generateLabel.textContent = isImage ? 'Reconstruct 3D' : 'Generate 3D';
+ setBusy(false);
updateGenerateEnabled();
}
+ // The button shows exactly one of: the spinner, the photo label, the text label.
+ // Toggling pre-translated nodes keeps every state localized without the script
+ // ever writing copy of its own.
+ function setBusy(busy) {
+ els.spinner.hidden = !busy;
+ els.labelImage.hidden = busy || state.mode !== 'image';
+ els.labelText.hidden = busy || state.mode !== 'text';
+ }
els.tabImage.addEventListener('click', () => setMode('image'));
els.tabText.addEventListener('click', () => setMode('text'));
@@ -1210,7 +1233,7 @@
state.abort = new AbortController();
els.generate.disabled = true;
- els.generateLabel.innerHTML = '<span class="spinner"></span>';
+ setBusy(true);
showState('loading');
startTimers();
@@ -1235,7 +1258,7 @@
} finally {
stopTimers();
els.generate.disabled = false;
- els.generateLabel.textContent = state.mode === 'image' ? 'Reconstruct 3D' : 'Generate 3D';
+ setBusy(false);
updateGenerateEnabled();
}
}
diff --git a/prompts/swarm-100/route-forge-nim.md b/prompts/swarm-100/route-forge-nim.md
deleted file mode 100644
index 8bab11708..000000000
--- a/prompts/swarm-100/route-forge-nim.md
+++ /dev/null
@@ -1,119 +0,0 @@
-# Route audit: /forge-nim (Forge · NIM: Self-Hosted Image to 3D)
-
-How to run: paste this file's repo path into a fresh Claude Code chat in this
-repository and say "run this work order". This file is fully self-contained:
-it depends on no other prompt file anywhere. If neighboring files in
-prompts/swarm-100/ are gone, that work is done; if present, ignore them.
-Every claim below rots; step 0 re-measures, and what you measure wins.
-
-## Operating clause (binding)
-
-- Read CLAUDE.md first. Its rules override everything, including this file.
-- Finish 100% in this session. Never end the turn with a question, an option
- list, or an unexecuted plan. A judgment call goes in one line of the final
- report; it never becomes a question that halts work.
-- The only permitted stops are the CLAUDE.md stop-and-ask gates: spending real
- funds or any irreversible on-chain write, git push or a production deploy,
- committing content that references a crypto project other than $THREE, and
- destroying unrecoverable data.
-- No mocks, no fake data, no placeholder stubs, no unfinished-work markers, no
- commented-out code. Real APIs and real integrations only.
-- The em-dash and en-dash characters are banned in everything you write.
-- Concurrent agents share this worktree: stage explicit paths only (never a
- bare add-everything), and commit finished work promptly.
-- Before committing, run: npm run check:rules -- --paths <files you touched>.
- It must exit 0.
-
-## Step 0: re-derive the current state
-
-1. This route's declared entry (path, description, added date):
-
- node -e "const p=require('./data/pages.json');for(const s of p.sections)for(const g of s.pages)if(g.path==='/forge-nim')console.log(JSON.stringify(g,null,2))"
-
-2. Find the serving implementation: grep "/forge-nim" in vercel.json (the live
- route table server/index.mjs boots from), then list matching files in
- pages/ and src/. STRUCTURE.md maps every surface to its directory if the
- grep is ambiguous.
-3. Production right now:
-
- curl -so /dev/null -w '%{http_code}\n' https://three.ws/forge-nim
-
-4. Start (or reuse) the local dev server: npm run dev (vite, port 3000).
-
-## Task: audit this one route end to end, then fix everything found
-
-Route: /forge-nim ("Forge · NIM: Self-Hosted Image to 3D", section "build").
-Declared purpose: Drive a self-hosted TRELLIS NIM (nvcr.io/nim/microsoft/trellis) directly: drop a photo, hit /v1/infer, and get the textured GLB back synchronously as base64: rendered live and downloadable.
-
-Drive the page in a real headless browser. Playwright is already a dev
-dependency; keep your probe script in the session scratchpad, never in the
-repo.
-
-1. Load http://localhost:3000/forge-nim. Record the HTTP status, every console
- error and warning, and every failed network request.
-2. Exercise it like a first-time user: click the primary actions, run the
- primary flow if one exists, follow the page's internal links. Every button
- must do something real; every link must resolve. If the page renders data,
- prove in the network log that it came from a live endpoint, not a baked-in
- fallback array.
-3. States: verify populated, empty, loading, and error states are designed and
- reachable. Block the page's primary fetch in Playwright and confirm an
- actionable error state appears, not a blank void or a raw exception.
-4. Responsive: render at 320, 768, and 1440 px. No horizontal page scroll, no
- overlapping or clipped controls.
-5. Accessibility basics: one h1; interactive elements are real buttons or
- links with visible focus states; images carry alt text; form fields have
- labels.
-6. Head: a specific title tag and meta description that match the declared
- purpose above.
-7. Fix every defect at its root in the page's real source files and re-run the
- probe until clean. Defects in shared modules are in scope when this route
- exercises them.
-
-## Definition of done
-
-- [ ] Local load of /forge-nim: HTTP 200, zero console errors, zero unhandled
- failed requests (a third-party outage must land in a designed fallback,
- not a broken page).
-- [ ] No dead interactive element on the page: no stub href="#", no button
- without a handler, no link to a missing path.
-- [ ] Empty, loading, error, and populated states verified.
-- [ ] 320 / 768 / 1440 px renders verified clean.
-- [ ] Title and meta description present and specific.
-- [ ] npm test passes; npm run check:rules -- --paths <touched files> exits 0.
-- [ ] Production status from step 0 recorded in the report; if production
- needs a deploy to pick up a fix, the report says so (deploys stay
- owner-gated).
-
-## Never blocked
-
-| Blocker | Resolution (act, do not ask) |
-|---|---|
-| Missing env var or credential | Check .env and .env.local, then the Cloud Run service env, then Secret Manager (the CLAUDE.md self-unblock playbook has the exact commands). If it truly exists nowhere, wire the code fully behind the env var, prove the wiring with a dry run, and name the single missing var in the report. |
-| gcloud not on PATH | Run: export PATH="$HOME/google-cloud-sdk/bin:$PATH" |
-| Dev server port 3000 busy | Another agent may be serving this repo; probe it and reuse if so. Otherwise start your own on a free port: npx vite --port 3101 |
-| Playwright browser missing | npx playwright install chromium |
-| A surface needs a signed-in user | Register a fresh account through the real /register flow against the real API and use it. Never mock the session. |
-| A defect sits in code you did not touch | Fix it if it blocks a Definition of done line (root cause it, never mask it). Otherwise note it in the report and continue. |
-| An unrelated test is red | Same rule. Never pipe npm test through tail; it masks exit codes. |
-
-## Close out (required)
-
-1. Verify every Definition of done line with the actual command output in
- front of you. Never claim a line you did not verify.
-2. Commit with explicit paths and a subject that describes the diff (house
- style: type(scope): what changed and why a reader cares).
-3. Delete this prompt file in that same commit:
-
- git rm prompts/swarm-100/route-forge-nim.md
-
- The shrinking directory is this campaign's only progress ledger; there is
- no progress log to update.
-4. Final report: what step 0 measured, what changed (file list), evidence per
- Definition of done line, and any one-line judgment calls. No trailing
- questions, no unexecuted plans.
-
-If a line genuinely cannot pass inside this session (an external party must
-respond, or an owner-gated action is the final step), finish everything else,
-leave this file in place, and state exactly which line remains and who owns
-it. Never delete this file on a partial.
diff --git a/public/changelog.json b/public/changelog.json
index 6dada7d70..1ba5b22fe 100644
--- a/public/changelog.json
+++ b/public/changelog.json
@@ -1,5 +1,5 @@
{
- "generated_at": "2026-08-17T05:22:51.398Z",
+ "generated_at": "2026-08-17T05:33:50.826Z",
"generated_by": "scripts/build-page-index.mjs from data/pages.json + data/changelog.json",
"site": {
"name": "three.ws",
@@ -16,6 +16,17 @@
"launch"
]
},
+ {
+ "date": "2026-08-17",
+ "type": "update",
+ "title": "A busy Forge engine now says it is busy and counts you back in, instead of telling you to go set an environment variable",
+ "summary": "The /forge page was driven end to end in a real browser. When every free 3D engine was momentarily at capacity, or a worker was still warming up, the server said so clearly and told the page to try again in about half a minute. The page threw that away. It treated any temporary outage as proof that the whole generator had never been set up, and showed a panel headed Generator not configured that asked you to set NVIDIA_API_KEY or HF_TOKEN on the deployment. For anyone who is not running the servers, which is everyone using the site, that is an instruction you cannot act on, about a problem you do not have, and the panel had no Try again button at all, so a wait of thirty seconds became a dead end that looked like the product was broken. A temporary outage now lands in the wait state the site already had for this: a heading that says Temporarily unavailable, the countdown ticking down to when the engines free up, and a Retry button that switches itself on the moment the wait is over. The genuine case, a deployment with no engine wired at all, still gets its original panel, because for the operator that message is the correct one. The engine picker also had a key underneath it explaining that a red dot means an engine is unavailable, but no engine could ever show that dot: only the amber busy dot was ever drawn, so a down engine was dimmed with no indication why, and the key described something that did not exist. Down engines now carry the red dot the key promises, at full strength so it stays distinct from the amber one. Separately, roughly one in ten preview thumbnails failed to load while developing locally, which made the gallery look like a broken image host when nothing was actually wrong with it.",
+ "link": "/forge",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
{
"date": "2026-08-17",
"type": "update",
@@ -268,6 +279,17 @@
"improvement"
]
},
+ {
+ "date": "2026-08-17",
+ "type": "update",
+ "title": "Skill tags now actually filter, a failed skills load says so instead of claiming the catalog is empty, and a shared skills link brings its filters with it",
+ "summary": "The /skills path was driven end to end in a real browser. It forwards into the Marketplace Skills tab, and it was forwarding a search that the tab then threw away: a link like /skills?q=whale landed on the full unfiltered list with an empty search box. The Skills tab now reads its search, category, tag, and sort straight off the address bar, and writes them back as you change them, so the view you are looking at is the view you can bookmark, share, reload, and step back through. The tag chips on a skill were the clearest casualty of that gap: clicking one navigated to a list that looked exactly the same as before, because nothing was filtering by tag at all. Tags now filter for real in the catalog itself, the tab shows which tag it is filtering by, and one click clears it. When the catalog cannot be reached, the grid used to show the words No skills found with an invitation to publish one, which is the same screen an empty catalog would give you and offers no way to try again: the error state and its Retry button were being overwritten a fraction of a second after they rendered. A failed load now says it could not load skills and hands you the retry. The count under the heading was also being reverted to its placeholder by the translation pass a moment after it was written, so it never showed how many skills you were looking at; it does now, and it names your active category or tag. The empty state names the filters that emptied the list instead of guessing. For anyone arriving without JavaScript, including crawlers, /skills was a completely blank page with no link out of it; it now carries a real heading, an explanation, and a link into the Marketplace. The search box has a proper label, the free and paid chips announce which one is selected, and the page holds up from 320px to 1440px.",
+ "link": "/marketplace?tab=skills",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
{
"date": "2026-08-17",
"type": "update",
diff --git a/public/changelog.xml b/public/changelog.xml
index 760a47df9..1a1f55f40 100644
--- a/public/changelog.xml
+++ b/public/changelog.xml
@@ -14,6 +14,14 @@
<category>launch</category>
<description>The Trench Car is the default car in every coin world: ambient traffic drives it and it is parked for you to take the wheel. How to drive, how a car is assembled from spec, physics and mesh, how the model is staged, and how to add another.</description>
</item>
+ <item>
+ <title>A busy Forge engine now says it is busy and counts you back in, instead of telling you to go set an environment variable</title>
+ <link>https://three.ws/forge</link>
+ <guid isPermaLink="false">2026-08-17:A busy Forge engine now says it is busy and counts you back in, instead of telling you to go set an environment variable</guid>
+ <pubDate>Mon, 17 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>The /forge page was driven end to end in a real browser. When every free 3D engine was momentarily at capacity, or a worker was still warming up, the server said so clearly and told the page to try again in about half a minute. The page threw that away. It treated any temporary outage as proof that the whole generator had never been set up, and showed a panel headed Generator not configured that asked you to set NVIDIA_API_KEY or HF_TOKEN on the deployment. For anyone who is not running the servers, which is everyone using the site, that is an instruction you cannot act on, about a problem you do not have, and the panel had no Try again button at all, so a wait of thirty seconds became a dead end that looked like the product was broken. A temporary outage now lands in the wait state the site already had for this: a heading that says Temporarily unavailable, the countdown ticking down to when the engines free up, and a Retry button that switches itself on the moment the wait is over. The genuine case, a deployment with no engine wired at all, still gets its original panel, because for the operator that message is the correct one. The engine picker also had a key underneath it explaining that a red dot means an engine is unavailable, but no engine could ever show that dot: only the amber busy dot was ever drawn, so a down engine was dimmed with no indication why, and the key described something that did not exist. Down engines now carry the red dot the key promises, at full strength so it stays distinct from the amber one. Separately, roughly one in ten preview thumbnails failed to load while developing locally, which made the gallery look like a broken image host when nothing was actually wrong with it.</description>
+ </item>
<item>
<title>A failed X seed no longer costs you six hours, and can no longer wipe a good one</title>
<link>https://three.ws/settings</link>
@@ -198,6 +206,14 @@
<category>fix</category>
<description>Five fixes to Selfie to Avatar. If your avatar library was already at the limit for your plan, nothing noticed until the very end: you waited out the full reconstruction, and then the page said the 3D engine had hit a snag and told you to try again. It was not an engine problem, trying again could never work, and every attempt spent another ninety seconds of generation to land in the same place. Your library is now checked the moment you press the button, so a full one is refused immediately with the actual reason and what to do about it, and a build that fills the last slot while it is running ends with that same plain explanation instead of the engine message. The page also stopped losing its own error messages: the Build button floats above the page, so tapping it from higher up left the explanation somewhere below the fold while the button quietly reset itself, which read as nothing having happened. Errors now scroll into view. On the finished-avatar screen, Learn how to embed pointed at a page that did not exist and answered 404; it now opens the embedding guide. And on a narrow phone the Photoreal and Stylized choices were squeezed into two columns barely wide enough for one word per line, so they now stack and can be read.</description>
</item>
+ <item>
+ <title>Skill tags now actually filter, a failed skills load says so instead of claiming the catalog is empty, and a shared skills link brings its filters with it</title>
+ <link>https://three.ws/marketplace?tab=skills</link>
+ <guid isPermaLink="false">2026-08-17:Skill tags now actually filter, a failed skills load says so instead of claiming the catalog is empty, and a shared skills link brings its filters with it</guid>
+ <pubDate>Mon, 17 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>The /skills path was driven end to end in a real browser. It forwards into the Marketplace Skills tab, and it was forwarding a search that the tab then threw away: a link like /skills?q=whale landed on the full unfiltered list with an empty search box. The Skills tab now reads its search, category, tag, and sort straight off the address bar, and writes them back as you change them, so the view you are looking at is the view you can bookmark, share, reload, and step back through. The tag chips on a skill were the clearest casualty of that gap: clicking one navigated to a list that looked exactly the same as before, because nothing was filtering by tag at all. Tags now filter for real in the catalog itself, the tab shows which tag it is filtering by, and one click clears it. When the catalog cannot be reached, the grid used to show the words No skills found with an invitation to publish one, which is the same screen an empty catalog would give you and offers no way to try again: the error state and its Retry button were being overwritten a fraction of a second after they rendered. A failed load now says it could not load skills and hands you the retry. The count under the heading was also being reverted to its placeholder by the translation pass a moment after it was written, so it never showed how many skills you were looking at; it does now, and it names your active category or tag. The empty state names the filters that emptied the list instead of guessing. For anyone arriving without JavaScript, including crawlers, /skills was a completely blank page with no link out of it; it now carries a real heading, an explanation, and a link into the Marketplace. The search box has a proper label, the free and paid chips announce which one is selected, and the page holds up from 320px to 1440px.</description>
+ </item>
<item>
<title>Talking Avatar Video stops handing the renderer a 3D file where a photo belongs, and says up front when the GPU lane is down</title>
<link>https://three.ws/create/video</link>
diff --git a/public/features.json b/public/features.json
index fc488fb95..8df1fdb93 100644
--- a/public/features.json
+++ b/public/features.json
@@ -1,5 +1,5 @@
{
- "generated_at": "2026-08-17T05:22:51.107Z",
+ "generated_at": "2026-08-17T05:33:50.722Z",
"generated_by": "scripts/build-page-index.mjs from data/pages.json",
"site": {
"name": "three.ws",
diff --git a/public/locales/en.json b/public/locales/en.json
index 14a01000d..bc897d815 100644
--- a/public/locales/en.json
+++ b/public/locales/en.json
@@ -7885,7 +7885,9 @@
"set_endpoint": "Set the endpoint",
"link_hosted_forge": "Use the hosted forge instead",
"link_nvidia_docs": "NVIDIA model docs",
- "retry_check": "Retry check"
+ "retry_check": "Retry check",
+ "label_reconstruct": "Reconstruct 3D",
+ "label_generate": "Generate 3D"
},
"forge_spark": {
"1_nemotron_refines_your_prompt_idle": "<span class=\"num\">1</span> <span><span class=\"who\">Nemotron</span><br><span class=\"does\">refines your prompt</span></span> <span class=\"state\">idle</span>",
@@ -8061,9 +8063,13 @@
},
"forgot_password": {
"back_to_sign_in": "<a href=\"/login\">Back to sign in</a>",
+ "couldn_t_reach_the_server_check_your": "Couldn't reach the server. Check your connection and try again.",
"email": "Email",
+ "enter_a_valid_email_address_datahintinvalid": "Enter a valid email address.",
"enter_your_email_and_we_ll_send": "Enter your email and we'll send you a reset link.",
"forgot_password": "Forgot password",
+ "if_an_account_exists_for_that_email": "If an account exists for that email, a reset link has been sent.",
+ "looks_valid_datahintvalid": "Looks valid",
"meta_description": "Request a password-reset email.",
"meta_og_description": "Request a password-reset email.",
"meta_og_title": "Forgot password · three.ws",
@@ -8071,6 +8077,8 @@
"meta_twitter_description": "Request a password-reset email.",
"meta_twitter_title": "Forgot password · three.ws",
"send_reset_link": "Send reset link",
+ "sending_datasendinglabel": "Sending…",
+ "something_went_wrong_on_our_end_and": "Something went wrong on our end and your reset link was not sent. Please try again.",
"three_ws": "three.ws",
"three_ws_home_arialabel": "three.ws home",
"we_ll_send_you_a_link_to": "We'll send you a link to reset your password."