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.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
docs(changelog): regenerate the changelog, features, llms and sitemap feeds for /duels and the new entries
nirholas636265f41831
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex cd39ce8cc..e68e9f117 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -7,12 +7,14 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr## 2026-09-30- **Platform Analytics** (`/analytics`): Every three.ws platform total on one page, read live: agents, agents with a wallet, coins launched, 3D models generated, LLM tokens processed, x402 settlements and volume, marketplace sales and agent-to-agent hire volume, each with a daily growth chart over 30 days, 90 days or all time and a note on exactly how it is counted.-- **Success Stories** (`/stories`): Agents with results anyone can check: coins launched, creator fees and service income, computed by three.ws from launch records and settled payments and shown only when the agent's owner opts in. Grouped by category, each linking to the full Spotlight entry and the coin on Solscan.+- **Trader Duels: call the winner** (`/duels`): Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.- **3D models no longer arrive glued to a floor slab**: The 3D generators sometimes built a thin, wide plate under whatever you asked for: a teapot sitting on a black marble square, a telescope standing in the middle of a grey sheet many times its size. Every finished model is now checked for that plate and has it cut away before it is saved, while anything that really has a base, like a lamp foot, a plinth or a tripod, is left exactly as it was. (`/forge`) `[fix]`- **59 more pages now read in your language**: Pick any of the 85 languages in the language menu and 59 more pages now follow it, including the Airdrop Checker, Portfolio, Drops, Crews, Launch, the Agent Monitor and the rig and embed doctors. Titles, buttons, form labels, hints, error messages and the text shown when a page is shared all translate. Live numbers and statuses on those pages (counters, balances, scan results, play and pause buttons) keep updating in every language: the translation step now recognises a value the page has already filled in and leaves it alone, instead of putting a placeholder like "Loading" back over it. (`/airdrops`) `[improvement, fix]`- **Every platform total in one free API call**: GET /api/platform/analytics returns agents, agents with a wallet, coins launched, 3D models generated, LLM tokens processed, x402 settlements and volume, marketplace sales and agent-to-agent hire volume, each with a daily series over 30 days, 90 days or all time and a plain sentence saying exactly which rows were counted. A figure that cannot be read comes back empty with a named reason instead of as zero. It powers the new analytics page and is open to anyone building on three.ws. (`/analytics`) `[feature]`- **Photo to 3D keeps going when its main engines go down**: When a photo-to-3D job fails partway through, it already moves to the next engine on its own. Until now that only worked while another textured engine was up, so when the two main engines failed together last week, photo generations stopped completely. There is now one more rung: our own TripoSG engine takes over the job and delivers the shape of your model without textures, and the progress screen tells you when that happens. You get a model you can work with instead of an error, and whenever a textured engine is up, it is still used first. (`/image-to-3d`) `[fix, improvement]`+- **See what every agent has earned**: Agent profiles, avatar pages and coin pages now show an Earned card: lifetime creator fees from the coins the agent launched, in SOL with USD beside it, what has been claimed and what is still waiting, a per-coin breakdown and every recorded claim with its Solscan link, plus income from selling its skills and services. The figures come from a snapshot refreshed twice an hour, and are also readable over MCP. `[feature]`- **See where every $100 paid in $THREE goes**: The $THREE page now shows, for paid compute, marketplace sales, scarcity mints, spins and copy-trading fees, how every $100 splits between the seller, the treasury that funds buybacks, and holder rewards. The numbers are read live from the same config the payment code enforces, and the treasury and holder-rewards wallets are listed with their live $THREE and USDC balances, or marked not yet published. The platform never burns $THREE. The page also follows your light or dark theme now. (`/three-token`) `[feature]`+- **Sell your whole agent as a paid API**: Flip one switch on your agent's Earn tab, set a price per call and a short description, and your agent becomes a pay-per-call API anyone can use. Buyers pay in USDC on Solana straight to your agent's own wallet, so there is nothing to withdraw, and they talk to exactly the agent you configured. Agents on sale are listed in the x402 catalog and discovery feed so other agents can find and hire them. `[feature]`- **The coming iPhone app gains notifications, a share sheet entry and quick actions**: The three.ws iOS app, on its way to the App Store, now matches the Android app and goes further. Turn on notifications and sales, follows and remixes arrive on your iPhone with your unread count on the app icon, and a tap opens the page it is about. three.ws appears in the share sheet: share one to three photos and they wait in Create as the front, left and right views of a new 3D avatar, or share a .glb from Files to upload it. Press and hold the icon for Create avatar, Discover, My agents and Notifications. (`/docs/ios-app`) `[feature]`- **The language menu now offers 69 languages**: The language menu grew from 14 languages to 69, each one fully translated. 55 more pages now follow the language you pick, including Hood Desk, Portfolios, Fee Bridge, Agent Cards, Connect, the CLI sign-in approval, Motion, Docs World and the legal pages. Their titles, headings, buttons and the text shown when a page is shared all translate, and so does the sitemap, so you can find every page on three.ws in your own language. (`/sitemap`) `[feature, improvement]`- **The Oracle's model now downloads from any website**: The Oracle's conviction model is published data, and the @three-ws/oracle-model package is meant to score coins anywhere. Until now its download endpoint only answered pages on three.ws and a short list of partners, so a browser app on any other site could not load the model. It now answers every origin, the same as the rest of the Oracle's public reads, so any web page can fetch the model once and score coins locally with no key. (`/oracle-lab`) `[improvement]`diff --git a/api/_lib/env.js b/api/_lib/env.jsindex 70bd5a063..6e8ebbdf6 100644--- a/api/_lib/env.js+++ b/api/_lib/env.js@@ -1536,6 +1536,32 @@ export const env = {.filter(Boolean);},+ // Additional NVIDIA NIM keys (build.nvidia.com), comma-separated. Each free+ // nvapi key carries its own rate limit, so the chain rotates across+ // NVIDIA_API_KEY then these before leaving the NIM lane.+ get NVIDIA_FALLBACK_KEYS() {+ return (opt('NVIDIA_FALLBACK_KEYS') || '')+ .split(',')+ .map((k) => k.trim())+ .filter(Boolean);+ },++ // Hugging Face token. The forge's free Spaces lane reads it for ZeroGPU+ // quota, and the LLM chain uses it for Inference Providers+ // (router.huggingface.co), which bills the account's monthly free credits.+ get HF_TOKEN() {+ return opt('HF_TOKEN');+ },++ // Additional Hugging Face tokens, comma-separated, each on its own account's+ // monthly Inference Providers credit. Tried after HF_TOKEN runs dry (402).+ get HF_FALLBACK_TOKENS() {+ return (opt('HF_FALLBACK_TOKENS') || '')+ .split(',')+ .map((k) => k.trim())+ .filter(Boolean);+ },+// Per-user daily LLM spend cap, in whole USD, for completions that route to// the platform's paid keys (Anthropic / OpenAI / Grok). Consumed by// _lib/llm.js (dailyCapMicroUsd) as a float; unset or non-positive falls backdiff --git a/api/_lib/llm-pricing.js b/api/_lib/llm-pricing.jsindex d56b32f8e..049dd5b7d 100644--- a/api/_lib/llm-pricing.js+++ b/api/_lib/llm-pricing.js@@ -104,6 +104,9 @@ const FREE_PROVIDERS = new Set(['cloudflare','siliconflow','llm7',+ // Inference Providers on each token's monthly included credit: the chain+ // never buys pre-paid credit, so an exhausted token 402s and fails over.+ 'huggingface',]);// Strip a rung suffix from a provider name: '#n' for multi-key rungsdiff --git a/api/_lib/llm.js b/api/_lib/llm.jsindex f7bd538f4..4526ab533 100644--- a/api/_lib/llm.js+++ b/api/_lib/llm.js@@ -88,6 +88,22 @@ const CEREBRAS_MODEL = 'llama-3.3-70b';// retired, so the composed id 404'd on every call while the base id looked fine.// Keep this in step with DEFAULT_FREE_MODEL in _lib/chat-models.js.const OPENROUTER_FREE_MODEL = DEFAULT_FREE_MODEL;+// Further `:free` routes, each on its own upstream pool. A saturated free model+// 429s every key at once ("Provider returned error"), so a second and third+// model, not a fourth key, is what keeps the lane serving. Verified live+// 2026-09-30: all three answered plain text and tool calls while one of them+// was 429ing. Every id keeps the ':free' suffix so the lane can never bill.+const OPENROUTER_ALT_FREE_MODELS = Object.freeze([+ 'nvidia/nemotron-3-super-120b-a12b:free',+ 'qwen/qwen3.8-27b:free',+]);+// Hugging Face Inference Providers (router.huggingface.co): OpenAI-compatible,+// routed to whichever partner serves the model, billed to the token's monthly+// included credit. Llama 3.3 70B answered clean text and JSON in ~0.5s under a+// 32-token budget (2026-09-30); gpt-oss on this router spends small budgets on+// reasoning and returns no content, so it is not used here.+const HF_ROUTER_URL = 'https://router.huggingface.co/v1/chat/completions';+const HF_MODEL = 'meta-llama/Llama-3.3-70B-Instruct';// Gemini Flash-Lite on the AI Studio FREE tier (GEMINI_API_KEY): an external// free quota, so it stays on the cheapest model.const GEMINI_MODEL = 'gemini-2.5-flash-lite';@@ -168,6 +184,63 @@ const POLLINATIONS_MODEL = 'openai-fast';// fails over to the reliable Vertex anchor in seconds. Read per-call (not a// load-time const) so it's tunable via env without a redeploy; floored so a bad// value can't disable the guard.+// Cross-request rung cooldown, per server instance. A rung whose key is out of+// credit (402), rejected (401/403) or whose model is retired (404/410) will give+// the same answer for a long while; a 429 for a short one. Without memory every+// completion re-asked each dead rung, and with twenty-odd free rungs on a bad+// day those round trips added up before the first live lane was reached.+// Keyed by rung name + model (the name already encodes which key). State lives+// on a global symbol so the test setup can clear it between tests without+// importing this module ahead of each file's mocks.+const COOLDOWNS = (globalThis[Symbol.for('three.ws.llmCooldowns')] ??= new Map());+const COOLDOWN_DEAD_MS = 30 * 60_000;+const COOLDOWN_RATE_DEFAULT_MS = 30_000;+const COOLDOWN_RATE_MAX_MS = 5 * 60_000;++function rungKey(p) {+ return `${p.name}|${p.model}`;+}++/**+ * How long a rung sits out after an HTTP failure, or 0 for "retry next call".+ * 5xx and transport errors are treated as transient and never cool a rung.+ */+export function cooldownMsForStatus(status, retryAfter = null) {+ if (status === 401 || status === 402 || status === 403 || status === 404 || status === 410) return COOLDOWN_DEAD_MS;+ if (status === 429) {+ const secs = Number(retryAfter);+ const ms = Number.isFinite(secs) && secs > 0 ? secs * 1000 : COOLDOWN_RATE_DEFAULT_MS;+ return Math.min(ms, COOLDOWN_RATE_MAX_MS);+ }+ return 0;+}++function noteRungFailure(p, status, retryAfter) {+ const ms = cooldownMsForStatus(status, retryAfter);+ if (ms > 0) COOLDOWNS.set(rungKey(p), { until: Date.now() + ms, status });+}++function rungCooling(p, now = Date.now()) {+ const c = COOLDOWNS.get(rungKey(p));+ if (!c) return false;+ if (c.until <= now) {+ COOLDOWNS.delete(rungKey(p));+ return false;+ }+ return true;+}++/** Rungs currently sitting out, for health endpoints: [{ rung, status, seconds }]. */+export function llmCooldownSnapshot(now = Date.now()) {+ return [...COOLDOWNS.entries()]+ .filter(([, c]) => c.until > now)+ .map(([rung, c]) => ({ rung, status: c.status, seconds: Math.ceil((c.until - now) / 1000) }));+}++export function resetLlmCooldowns() {+ COOLDOWNS.clear();+}+// Per-provider wall clock for one rung of the chain. Shared so a rung that wants// "the same budget as everyone else" cannot drift from the number llmComplete// actually enforces.@@ -239,7 +312,10 @@ export async function checkUserLlmSpendCap(userId, { anthropicKey, grokKey } = {-- on the platform key is real spend and counts against the cap.AND NOT (provider LIKE 'openrouter%' AND (model IS NULL OR model LIKE '%:free'))AND provider NOT LIKE 'vertex%'- AND provider NOT IN ('nvidia', 'cerebras', 'gemini', 'ovh', 'pollinations', 'sambanova', 'mistral', 'zai', 'cloudflare', 'siliconflow', 'llm7')+ -- Multi-key rungs (nvidia#2, huggingface#2) meter like their base lane.+ AND provider NOT LIKE 'nvidia%'+ AND provider NOT LIKE 'huggingface%'+ AND provider NOT IN ('cerebras', 'gemini', 'ovh', 'pollinations', 'sambanova', 'mistral', 'zai', 'cloudflare', 'siliconflow', 'llm7')AND created_at > NOW() - INTERVAL '24 hours'`;const spent = Number(row?.spent ?? 0);@@ -571,10 +647,23 @@ export function providerChain({ anthropicKey, anthropicModel, grokKey = null, grmodel: OPENROUTER_FREE_MODEL,}));});- if (env.NVIDIA_API_KEY) {+ // The alternate free models, one rung each, rotating across the keys so a+ // single account's daily free-request cap never carries them all.+ if (openrouterKeys.length) {+ OPENROUTER_ALT_FREE_MODELS.filter((m) => m !== OPENROUTER_FREE_MODEL).forEach((model, i) => {+ chain.push(openrouterProvider({+ name: `openrouter:${model.split('/')[1].replace(/:free$/, '')}`,+ key: openrouterKeys[i % openrouterKeys.length],+ model,+ }));+ });+ }+ // One NIM rung per nvapi key: each free key carries its own rate limit.+ const nvidiaKeys = [...new Set([env.NVIDIA_API_KEY, ...env.NVIDIA_FALLBACK_KEYS].filter(Boolean))];+ nvidiaKeys.forEach((key, i) => {chain.push(openaiCompatProvider({- name: 'nvidia',- key: env.NVIDIA_API_KEY,+ name: i === 0 ? 'nvidia' : `nvidia#${i + 1}`,+ key,url: 'https://integrate.api.nvidia.com/v1/chat/completions',model: NVIDIA_MODEL,extraBody: NVIDIA_NO_THINK,@@ -584,7 +673,7 @@ export function providerChain({ anthropicKey, anthropicModel, grokKey = null, gr// hardcoded 6s.timeoutMs: nvidiaLaneTimeoutMs(),}));- }+ });// SambaNova's free tier: the same Llama 3.3 70B on a fourth independent// quota pool, so a day that exhausts Groq, Cerebras, and NVIDIA at once// still has a 70B-class free rung with budget left.@@ -627,6 +716,19 @@ export function providerChain({ anthropicKey, anthropicModel, grokKey = null, grmodel: CLOUDFLARE_MODEL,}));}+ // Hugging Face Inference Providers, one rung per token. A free account's+ // monthly credit is small (a dozen short calls drained a fresh one), so+ // these sit behind the lanes with daily quotas; an exhausted token answers+ // 402 in ~100ms and the provider cooldown below keeps it out of later calls.+ const hfTokens = [...new Set([env.HF_TOKEN, ...env.HF_FALLBACK_TOKENS].filter(Boolean))];+ hfTokens.forEach((key, i) => {+ chain.push(openaiCompatProvider({+ name: i === 0 ? 'huggingface' : `huggingface#${i + 1}`,+ key,+ url: HF_ROUTER_URL,+ model: HF_MODEL,+ }));+ });// OVH AI Endpoints anonymous tier — no key required, always available.// Last of the 70B-class free rungs because its per-model anonymous quota// (2 req/min/IP) is the tightest in the chain; everything with a real key@@ -829,7 +931,14 @@ export async function llmComplete({ system, user, maxTokens = 1024, anthropicKeyconst stalledHosts = new Set();const hostOf = (url) => { try { return new URL(url).host; } catch { return url; } };const isStall = (e) => e?.name === 'TimeoutError' || e?.name === 'AbortError' || /abort|timed? ?out/i.test(e?.message || '');+ // Honor cooldowns only while something else can still answer: when every+ // rung is sitting out, try them all anyway rather than fail without asking.+ const everyRungCooling = chain.every((p) => rungCooling(p));for (const p of chain) {+ if (!everyRungCooling && rungCooling(p)) {+ attempts.push({ provider: p.name, skipped: 'cooling' });+ continue;+ }const remaining = deadline - Date.now();// Out of overall budget — stop rather than start an attempt we can't finish.if (remaining <= 500) {@@ -876,6 +985,7 @@ export async function llmComplete({ system, user, maxTokens = 1024, anthropicKey}if (!upstream.ok) {const body = await upstream.text().catch(() => '');+ noteRungFailure(p, upstream.status, upstream.headers?.get?.('retry-after'));lastErr = Object.assign(new Error(`${p.name} ${upstream.status}: ${body.slice(0, 200)}`), { status: 502, code: 'upstream_error' });attempts.push({ provider: p.name, ms: Date.now() - startedAt, error: `http ${upstream.status}` });continue;@@ -907,6 +1017,7 @@ export async function llmComplete({ system, user, maxTokens = 1024, anthropicKeyattempts.push({ provider: p.name, ms: Date.now() - startedAt, error: 'empty completion' });continue;}+ COOLDOWNS.delete(rungKey(p));recordLlmSpend(p, usage, Date.now() - startedAt, track);return {text,diff --git a/api/_lib/vision.js b/api/_lib/vision.jsindex d43afab3e..5e6b6e252 100644--- a/api/_lib/vision.js+++ b/api/_lib/vision.js@@ -117,20 +117,25 @@ const NVIDIA_VISION_MODELS = [// Replace this ordering with measured latency and JSON-success rates by running// `node scripts/probe-vision-lanes.mjs`, which sends the real rubric payload to// every route and ranks them.+//+// Pruned 2026-09-30 against a live image probe: ling-3.0-flash-vl and both+// nex-agi routes lost their free tier (404 "unavailable for free") and both+// inkling routes are now restricted to OpenRouter's agentic surfaces (403), so+// every vision request was paying five dead round trips. qwen3.8-27b:free+// joined: the catalog lists it with image input.const OPENROUTER_VISION_MODELS = [{ model: 'google/gemma-4-31b-it:free' },{ model: 'google/gemma-4-26b-a4b-it:free' },- { model: 'inclusionai/ling-3.0-flash-vl:free', extraBody: { reasoning: { effort: 'none' } } },+ { model: 'qwen/qwen3.8-27b:free' },{ model: 'dots-studio/dots-3-note-preview:free' },- { model: 'nex-agi/nex-n2.5-pro:free' },- { model: 'nex-agi/nex-n2.5-mini:free' },- { model: 'thinkingmachines/inkling-small:free', extraBody: { reasoning: { effort: 'none' } } },- { model: 'thinkingmachines/inkling:free', extraBody: { reasoning: { effort: 'none' } } },{ model: 'nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free', extraBody: { reasoning: { effort: 'none' } } },];-// Cloudflare Workers AI's vision route. Same weights as the NIM lane above, so-// it is a capacity and quota hedge rather than a new capability.-const CLOUDFLARE_VISION_MODEL = '@cf/meta/llama-3.2-11b-vision-instruct';+// Cloudflare Workers AI's vision route. Llama 4 Scout, not the 11B Llama 3.2+// vision model the NIM lane uses: on Workers AI that one sits behind a Meta+// license click-through (403 "Model Agreement") that has to be accepted on the+// account by its owner, while Scout answered a real image correctly in ~2s+// with no gate (probed 2026-09-30).+const CLOUDFLARE_VISION_MODEL = '@cf/meta/llama-4-scout-17b-16e-instruct';// Paid last-resort tail. gpt-5.4-nano is vision-capable and already priced in// llm-pricing.js, keeping the backstop cheap and the spend ledger truthful.const OPENAI_VISION_MODEL = 'gpt-5.4-nano';@@ -306,31 +311,36 @@ function openaiCompatVisionProvider({ name, key, url, model, getHeaders = null,// (tests/api/llm-vertex-anchor-surfaces).export function visionChain() {const chain = [];- if (env.NVIDIA_API_KEY) {+ // One lane per nvapi key: each free key carries its own rate limit.+ const nvidiaKeys = [...new Set([env.NVIDIA_API_KEY, ...env.NVIDIA_FALLBACK_KEYS].filter(Boolean))];+ nvidiaKeys.forEach((key, i) => {for (const model of NVIDIA_VISION_MODELS) {chain.push(openaiCompatVisionProvider({- name: 'nvidia',- key: env.NVIDIA_API_KEY,+ name: i === 0 ? 'nvidia' : `nvidia#${i + 1}`,+ key,url: 'https://integrate.api.nvidia.com/v1/chat/completions',model,}));}- }+ });// Free OpenRouter routes, alongside the NIM lanes and ahead of the credits// anchor: same free tier, same zero meter, and they are the rungs that keep// vision answering when both NVIDIA and Google are down at once. Ordering// them here leaves the Vertex anchor exactly where it was in the chain// relative to the paid tail; nothing is evicted.- if (env.OPENROUTER_API_KEY) {- for (const spec of OPENROUTER_VISION_MODELS) {+ // The routes rotate across every OpenRouter key, so one account's daily+ // free-request cap never carries the whole vision lane.+ const openrouterKeys = [...new Set([env.OPENROUTER_API_KEY, ...env.OPENROUTER_FALLBACK_KEYS].filter(Boolean))];+ if (openrouterKeys.length) {+ OPENROUTER_VISION_MODELS.forEach((spec, i) => {chain.push(openaiCompatVisionProvider({name: 'openrouter',- key: env.OPENROUTER_API_KEY,+ key: openrouterKeys[i % openrouterKeys.length],url: 'https://openrouter.ai/api/v1/chat/completions',model: spec.model,extraBody: spec.extraBody || null,}));- }+ });}// Cloudflare Workers AI: the SAME llama-3.2-11b-vision weights the free NIM// lane already answers with, served on a different host behind a differentdiff --git a/public/changelog-recent.json b/public/changelog-recent.jsonindex 0148cf3e7..bc42c4776 100644--- a/public/changelog-recent.json+++ b/public/changelog-recent.json@@ -1,11 +1,11 @@{- "generated_at": "2026-09-30T08:27:19.820Z",+ "generated_at": "2026-09-30T08:30:15.454Z","generated_by": "scripts/build-page-index.mjs from data/pages.json + data/changelog.json","site": {"name": "three.ws","url": "https://three.ws"},- "total": 3790,+ "total": 3792,"complete": false,"full_feed": "/changelog.json","entries": [@@ -22,9 +22,9 @@{"date": "2026-09-30","type": "launch",- "title": "Success Stories",- "summary": "Agents with results anyone can check: coins launched, creator fees and service income, computed by three.ws from launch records and settled payments and shown only when the agent's owner opts in. Grouped by category, each linking to the full Spotlight entry and the coin on Solscan.",- "link": "/stories",+ "title": "Trader Duels: call the winner",+ "summary": "Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.",+ "link": "/duels","tags": ["launch"]@@ -71,6 +71,16 @@"improvement"]},+ {+ "date": "2026-09-30",+ "type": "update",+ "title": "See what every agent has earned",+ "summary": "Agent profiles, avatar pages and coin pages now show an Earned card: lifetime creator fees from the coins the agent launched, in SOL with USD beside it, what has been claimed and what is still waiting, a per-coin breakdown and every recorded claim with its Solscan link, plus income from selling its skills and services. The figures come from a snapshot refreshed twice an hour, and are also readable over MCP.",+ "link": null,+ "tags": [+ "feature"+ ]+ },{"date": "2026-09-30","type": "update",@@ -81,6 +91,16 @@"feature"]},+ {+ "date": "2026-09-30",+ "type": "update",+ "title": "Sell your whole agent as a paid API",+ "summary": "Flip one switch on your agent's Earn tab, set a price per call and a short description, and your agent becomes a pay-per-call API anyone can use. Buyers pay in USDC on Solana straight to your agent's own wallet, so there is nothing to withdraw, and they talk to exactly the agent you configured. Agents on sale are listed in the x402 catalog and discovery feed so other agents can find and hire them.",+ "link": null,+ "tags": [+ "feature"+ ]+ },{"date": "2026-09-30","type": "update",@@ -1564,27 +1584,6 @@"tags": ["improvement"]- },- {- "date": "2026-09-09",- "type": "update",- "title": "Fade Radar: the wallets that have never once been right, and the coins they are crowding into",- "summary": "The Smart Money Radar tells you which proven winners are buying a coin. Fade Radar is the other half of that question, and on pump.fun it turns out to be the sharper one: how much of a coin's buy side is money that has never been right. A wallet joins the reverse-indicator board only after five coins it bought have a known outcome and not one of them won, which is a bar a surprising number of addresses clear. The new page ranks every launch from the last few hours by how much of its observed buy side belongs to those wallets, by wallet count and by SOL, and carries a second board of the wallets themselves with their full record. The number above the board is the part we care most about: it is measured, not asserted. We rebuild the reverse-indicator group using only coins whose outcome was known before the coins it is graded on existed, then check what actually happened. Coins where a quarter or more of the buyers were reverse indicators went on to win 1.8 percent of the time, against 22.1 percent for coins with none, across 26,626 labelled coins. Nothing here is a short and nothing signs anything: on a bonding curve the only tradeable form of a fade is not buying, so every action on the page is a link out to the coin, its intel, or a wallet's track record. Wallet links also now open directly on the Smart Money Radar, so an address cited anywhere on the platform lands on its record instead of a board you have to scan by eye.",- "link": "/fade",- "tags": [- "feature"- ]- },- {- "date": "2026-09-09",- "type": "update",- "title": "Homes connected through the relay now follow the house when you rearrange it",- "summary": "If your Home Assistant is only reachable inside your house, three.ws reaches it through the three.ws integration you install there, which dials out to us. Those homes were building their room list once, when the connection opened, and then quietly ignoring changes: assign a lamp to the kitchen, rename a floor, add a device, and the 3D house kept showing the old arrangement until the connection happened to be rebuilt. Homes connected by a direct address never had this problem, and the split was in the relay, which was not carrying the four small notices Home Assistant sends when its rooms, floors, devices or entities change. It carries them now, so both kinds of connection track the house the same way and updates land in seconds. The notices only say that something changed; the relay already read the full room and floor lists, so nothing new about your house is exposed by this.",- "link": "/smart-home",- "tags": [- "fix",- "infra"- ]}]}diff --git a/public/changelog.json b/public/changelog.jsonindex 3e99e8058..0b7205b28 100644--- a/public/changelog.json+++ b/public/changelog.json@@ -1,5 +1,5 @@{- "generated_at": "2026-09-30T08:19:59.945Z",+ "generated_at": "2026-09-30T08:30:15.416Z","generated_by": "scripts/build-page-index.mjs from data/pages.json + data/changelog.json","site": {"name": "three.ws",@@ -16,6 +16,16 @@"launch"]},+ {+ "date": "2026-09-30",+ "type": "launch",+ "title": "Trader Duels: call the winner",+ "summary": "Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.",+ "link": "/duels",+ "tags": [+ "launch"+ ]+ },{"date": "2026-09-30","type": "update",@@ -58,6 +68,16 @@"improvement"]},+ {+ "date": "2026-09-30",+ "type": "update",+ "title": "See what every agent has earned",+ "summary": "Agent profiles, avatar pages and coin pages now show an Earned card: lifetime creator fees from the coins the agent launched, in SOL with USD beside it, what has been claimed and what is still waiting, a per-coin breakdown and every recorded claim with its Solscan link, plus income from selling its skills and services. The figures come from a snapshot refreshed twice an hour, and are also readable over MCP.",+ "link": null,+ "tags": [+ "feature"+ ]+ },{"date": "2026-09-30","type": "update",@@ -68,6 +88,26 @@"feature"]},+ {+ "date": "2026-09-30",+ "type": "update",+ "title": "Sell your whole agent as a paid API",+ "summary": "Flip one switch on your agent's Earn tab, set a price per call and a short description, and your agent becomes a pay-per-call API anyone can use. Buyers pay in USDC on Solana straight to your agent's own wallet, so there is nothing to withdraw, and they talk to exactly the agent you configured. Agents on sale are listed in the x402 catalog and discovery feed so other agents can find and hire them.",+ "link": null,+ "tags": [+ "feature"+ ]+ },+ {+ "date": "2026-09-30",+ "type": "update",+ "title": "The coming iPhone app gains notifications, a share sheet entry and quick actions",+ "summary": "The three.ws iOS app, on its way to the App Store, now matches the Android app and goes further. Turn on notifications and sales, follows and remixes arrive on your iPhone with your unread count on the app icon, and a tap opens the page it is about. three.ws appears in the share sheet: share one to three photos and they wait in Create as the front, left and right views of a new 3D avatar, or share a .glb from Files to upload it. Press and hold the icon for Create avatar, Discover, My agents and Notifications.",+ "link": "/docs/ios-app",+ "tags": [+ "feature"+ ]+ },{"date": "2026-09-30","type": "update",diff --git a/public/changelog.xml b/public/changelog.xmlindex da30c32a1..7c5d2e6a0 100644--- a/public/changelog.xml+++ b/public/changelog.xml@@ -14,6 +14,14 @@<category>launch</category><description>Every three.ws platform total on one page, read live: agents, agents with a wallet, coins launched, 3D models generated, LLM tokens processed, x402 settlements and volume, marketplace sales and agent-to-agent hire volume, each with a daily growth chart over 30 days, 90 days or all time and a note on exactly how it is counted.</description></item>+ <item>+ <title>Trader Duels: call the winner</title>+ <link>https://three.ws/duels</link>+ <guid isPermaLink="false">2026-09-30:Trader Duels: call the winner</guid>+ <pubDate>Wed, 30 Sep 2026 12:00:00 GMT</pubDate>+ <category>launch</category>+ <description>Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.</description>+ </item><item><title>3D models no longer arrive glued to a floor slab</title><link>https://three.ws/forge</link>@@ -46,6 +54,14 @@<category>fix</category><description>When a photo-to-3D job fails partway through, it already moves to the next engine on its own. Until now that only worked while another textured engine was up, so when the two main engines failed together last week, photo generations stopped completely. There is now one more rung: our own TripoSG engine takes over the job and delivers the shape of your model without textures, and the progress screen tells you when that happens. You get a model you can work with instead of an error, and whenever a textured engine is up, it is still used first.</description></item>+ <item>+ <title>See what every agent has earned</title>+ <link>https://three.ws/changelog</link>+ <guid isPermaLink="false">2026-09-30:See what every agent has earned</guid>+ <pubDate>Wed, 30 Sep 2026 12:00:00 GMT</pubDate>+ <category>feature</category>+ <description>Agent profiles, avatar pages and coin pages now show an Earned card: lifetime creator fees from the coins the agent launched, in SOL with USD beside it, what has been claimed and what is still waiting, a per-coin breakdown and every recorded claim with its Solscan link, plus income from selling its skills and services. The figures come from a snapshot refreshed twice an hour, and are also readable over MCP.</description>+ </item><item><title>See where every $100 paid in $THREE goes</title><link>https://three.ws/three-token</link>@@ -54,6 +70,22 @@<category>feature</category><description>The $THREE page now shows, for paid compute, marketplace sales, scarcity mints, spins and copy-trading fees, how every $100 splits between the seller, the treasury that funds buybacks, and holder rewards. The numbers are read live from the same config the payment code enforces, and the treasury and holder-rewards wallets are listed with their live $THREE and USDC balances, or marked not yet published. The platform never burns $THREE. The page also follows your light or dark theme now.</description></item>+ <item>+ <title>Sell your whole agent as a paid API</title>+ <link>https://three.ws/changelog</link>+ <guid isPermaLink="false">2026-09-30:Sell your whole agent as a paid API</guid>+ <pubDate>Wed, 30 Sep 2026 12:00:00 GMT</pubDate>+ <category>feature</category>+ <description>Flip one switch on your agent's Earn tab, set a price per call and a short description, and your agent becomes a pay-per-call API anyone can use. Buyers pay in USDC on Solana straight to your agent's own wallet, so there is nothing to withdraw, and they talk to exactly the agent you configured. Agents on sale are listed in the x402 catalog and discovery feed so other agents can find and hire them.</description>+ </item>+ <item>+ <title>The coming iPhone app gains notifications, a share sheet entry and quick actions</title>+ <link>https://three.ws/docs/ios-app</link>+ <guid isPermaLink="false">2026-09-30:The coming iPhone app gains notifications, a share sheet entry and quick actions</guid>+ <pubDate>Wed, 30 Sep 2026 12:00:00 GMT</pubDate>+ <category>feature</category>+ <description>The three.ws iOS app, on its way to the App Store, now matches the Android app and goes further. Turn on notifications and sales, follows and remixes arrive on your iPhone with your unread count on the app icon, and a tap opens the page it is about. three.ws appears in the share sheet: share one to three photos and they wait in Create as the front, left and right views of a new 3D avatar, or share a .glb from Files to upload it. Press and hold the icon for Create avatar, Discover, My agents and Notifications.</description>+ </item><item><title>The language menu now offers 69 languages</title><link>https://three.ws/sitemap</link>diff --git a/public/features.json b/public/features.jsonindex 9ad150114..c1d771b91 100644--- a/public/features.json+++ b/public/features.json@@ -1,5 +1,5 @@{- "generated_at": "2026-09-30T08:19:59.918Z",+ "generated_at": "2026-09-30T08:30:15.396Z","generated_by": "scripts/build-page-index.mjs from data/pages.json","site": {"name": "three.ws",@@ -2186,6 +2186,16 @@"tags": [],"showcase": false},+ {+ "path": "/duels",+ "title": "Trader Duels: call the winner",+ "description": "Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.",+ "added": "2026-09-30",+ "auth": null,+ "indexable": true,+ "tags": [],+ "showcase": false+ },{"path": "/arena","title": "The Arena: live PvP trading tournaments",diff --git a/public/llms-full.txt b/public/llms-full.txtindex 0701f06ca..5d6557007 100644--- a/public/llms-full.txt+++ b/public/llms-full.txt@@ -1297,6 +1297,12 @@ Indexable: no (excluded from sitemap)Live command center for all 4 autonomous agent capabilities: Alpha Hunt smart-money scoring, autonomous Coin Launcher on schedule, Creator Auto-Claim for fee harvesting, and Market Maker with Jito execution.+### Trader Duels: call the winner++URL: https://three.ws/duels++Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.+### The Arena: live PvP trading tournamentsURL: https://three.ws/arenadiff --git a/public/llms.txt b/public/llms.txtindex 8f1340fe0..7617b4ff4 100644--- a/public/llms.txt+++ b/public/llms.txt@@ -95,6 +95,6 @@ Machine-readable entry points. Fetch these instead of scraping HTML.## Optional-- [llms-full.txt](https://three.ws/llms-full.txt): Every public page (916), grouped by section, with full descriptions.+- [llms-full.txt](https://three.ws/llms-full.txt): Every public page (917), grouped by section, with full descriptions.- [Sitemap](https://three.ws/sitemap): The same index for people, with search.- [sitemap.xml](https://three.ws/sitemap.xml): XML sitemap for crawlers, including every agent and avatar page.diff --git a/public/sitemap/index.html b/public/sitemap/index.htmlindex b7964f61d..46e99497f 100644--- a/public/sitemap/index.html+++ b/public/sitemap/index.html@@ -3,16 +3,16 @@<head><meta charset="utf-8" /><meta name="viewport" content="width=device-width,initial-scale=1" />-<title>Sitemap · three.ws</title>-<meta name="description" content="Human-readable index of every page on three.ws." />+<title data-i18n="sitemap_index.meta_title">Sitemap · three.ws</title>+<meta name="description" content="Human-readable index of every page on three.ws." data-i18n-attr="content:sitemap_index.meta_description"/><meta name="robots" content="index, follow, max-image-preview:large, max-snippet:-1" /><link rel="canonical" href="https://three.ws/sitemap" /><meta property="og:type" content="website" /><meta property="og:site_name" content="three.ws" /><meta property="og:locale" content="en_US" /><meta property="og:url" content="https://three.ws/sitemap" />-<meta property="og:title" content="Sitemap · three.ws" />-<meta property="og:description" content="Human-readable index of every page on three.ws." />+<meta property="og:title" content="Sitemap · three.ws" data-i18n-attr="content:sitemap_index.meta_og_title"/>+<meta property="og:description" content="Human-readable index of every page on three.ws." data-i18n-attr="content:sitemap_index.meta_og_description"/><meta property="og:image" content="https://three.ws/api/page-og?s=main&t=Sitemap&d=Human-readable+index+of+every+page+on+three.ws.&p=%2Fsitemap" /><meta property="og:image:width" content="1200" /><meta property="og:image:height" content="630" />@@ -20,8 +20,8 @@<meta property="og:image:alt" content="Sitemap - three.ws" /><meta name="twitter:card" content="summary_large_image" /><meta name="twitter:site" content="@trythreews" />-<meta name="twitter:title" content="Sitemap · three.ws" />-<meta name="twitter:description" content="Human-readable index of every page on three.ws." />+<meta name="twitter:title" content="Sitemap · three.ws" data-i18n-attr="content:sitemap_index.meta_twitter_title"/>+<meta name="twitter:description" content="Human-readable index of every page on three.ws." data-i18n-attr="content:sitemap_index.meta_twitter_description"/><meta name="twitter:image" content="https://three.ws/api/page-og?s=main&t=Sitemap&d=Human-readable+index+of+every+page+on+three.ws.&p=%2Fsitemap" /><script type="application/ld+json">{"@context":"https://schema.org","@graph":[{"@type":"WebPage","name":"Sitemap","description":"Human-readable index of every page on three.ws.","url":"https://three.ws/sitemap","isPartOf":{"@type":"WebSite","name":"three.ws","url":"https://three.ws"},"primaryImageOfPage":"https://three.ws/api/page-og?s=main&t=Sitemap&d=Human-readable+index+of+every+page+on+three.ws.&p=%2Fsitemap"}]}</script><link rel="alternate" type="application/xml" title="XML sitemap" href="/sitemap.xml" />@@ -95,27 +95,35 @@<header><div id="nav-container"></div></header><main class="sm-wrap"><div class="sm-hero">- <h1>Sitemap</h1>- <p><span>966</span> <span data-i18n="sitemap_index.pages_grouped">pages on three.ws, grouped by purpose. Looking for the machine-readable versions?</span></p>+ <h1 data-i18n="sitemap_index.sitemap">Sitemap</h1>+ <p><span>967</span> <span data-i18n="sitemap_index.pages_grouped">pages on three.ws, grouped by purpose. Looking for the machine-readable versions?</span></p><div class="sm-formats">- <a href="/sitemap.xml"><code>sitemap.xml</code> · for search engines</a>- <a href="/llms.txt"><code>llms.txt</code> · for AI agents</a>- <a href="/llms-full.txt"><code>llms-full.txt</code> · long form</a>- <a href="/api/features.json"><code>features.json</code> · machine manifest</a>- <a href="/changelog"><code>changelog</code> · what's new</a>- <a href="/openapi.json"><code>openapi.json</code> · HTTP API</a>+ <a href="/sitemap.xml" data-i18n-html="sitemap_index.for_search_engines"><code>sitemap.xml</code> · for search engines</a>+ <a href="/llms.txt" data-i18n-html="sitemap_index.for_ai_agents"><code>llms.txt</code> · for AI agents</a>+ <a href="/llms-full.txt" data-i18n-html="sitemap_index.long_form"><code>llms-full.txt</code> · long form</a>+ <a href="/api/features.json" data-i18n-html="sitemap_index.machine_manifest"><code>features.json</code> · machine manifest</a>+ <a href="/changelog" data-i18n-html="sitemap_index.what_s_new"><code>changelog</code> · what's new</a>+ <a href="/openapi.json" data-i18n-html="sitemap_index.http_api"><code>openapi.json</code> · HTTP API</a></div></div><div class="sm-filter"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" aria-hidden="true"><circle cx="11" cy="11" r="7"/><path d="m21 21-4.3-4.3"/></svg>- <input id="sm-filter-input" type="search" placeholder="Filter 966 pages by name, path, or description…" aria-label="Filter pages" autocomplete="off" spellcheck="false" />+ <input id="sm-filter-input" type="search" placeholder="Filter 967 pages by name, path, or description…" aria-label="Filter pages" autocomplete="off" spellcheck="false" /><kbd id="sm-filter-kbd" aria-hidden="true">/</kbd></div><p class="sm-filter-count" role="status" aria-live="polite"><span id="sm-filter-count"></span><span id="sm-filter-hint" aria-hidden="true"></span></p><section class="sm-new" id="sm-new" aria-labelledby="sm-new-title">- <h2 id="sm-new-title">Newest<a class="sm-new-more" href="/changelog">full changelog →</a></h2>- <p class="sm-section-desc">The 9 most recently launched pages, straight to the live surface.</p>+ <h2 id="sm-new-title" data-i18n-html="sitemap_index.newestfull_changelog">Newest<a class="sm-new-more" href="/changelog">full changelog →</a></h2>+ <p class="sm-section-desc" data-i18n="sitemap_index.the_9_most_recently_launched_pages_straight">The 9 most recently launched pages, straight to the live surface.</p><ul class="sm-list">+ <li>+ <a href="/duels">+ <span class="sm-new-date" translate="no">Sep 30, 2026</span>+ <span class="sm-title"><span data-i18n="sitemap_index.t_duels">Trader Duels: call the winner</span></span>+ <span class="sm-path" translate="no">/duels</span>+ <span class="sm-desc" data-i18n="sitemap_index.d_duels">Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.</span>+ </a>+ </li><li><a href="/analytics"><span class="sm-new-date" translate="no">Sep 30, 2026</span>@@ -180,33 +188,25 @@<span class="sm-desc" data-i18n="sitemap_index.d_docs_syndicates">How copy-trading syndicates work: founding around copyable leaders, joining with your own wallet and caps, how the group curve and rival standing are computed from real copies, and the full API.</span></a></li>- <li>- <a href="/docs/trading-quests">- <span class="sm-new-date" translate="no">Sep 29, 2026</span>- <span class="sm-title"><span data-i18n="sitemap_index.t_docs_trading_quests">Docs · Daily trading quests</span></span>- <span class="sm-path" translate="no">/docs/trading-quests</span>- <span class="sm-desc" data-i18n="sitemap_index.d_docs_trading_quests">The five daily trading quests, what proof each one is scored from, XP levels and quest badges, the disarmed $THREE reward switch, and the quests API.</span>- </a>- </li></ul></section><nav class="sm-toc" aria-label="Sections">- <a href="#main">Main</a>- <a href="#build">Build</a>- <a href="#labs">Labs</a>- <a href="#crypto">Crypto</a>- <a href="#agent-tools">Agent Tools</a>- <a href="#account">Account</a>- <a href="#learn">Learn</a>- <a href="#blog">Blog</a>- <a href="#legal">Legal</a>- <a href="#machine">Machine-readable</a>- <a href="#news">News</a>+ <a href="#main" data-i18n="sitemap_index.main">Main</a>+ <a href="#build" data-i18n="sitemap_index.build">Build</a>+ <a href="#labs" data-i18n="sitemap_index.labs">Labs</a>+ <a href="#crypto" data-i18n="sitemap_index.crypto">Crypto</a>+ <a href="#agent-tools" data-i18n="sitemap_index.agent_tools">Agent Tools</a>+ <a href="#account" data-i18n="sitemap_index.account">Account</a>+ <a href="#learn" data-i18n="sitemap_index.learn">Learn</a>+ <a href="#blog" data-i18n="sitemap_index.blog">Blog</a>+ <a href="#legal" data-i18n="sitemap_index.legal">Legal</a>+ <a href="#machine" data-i18n="sitemap_index.machine_readable">Machine-readable</a>+ <a href="#news" data-i18n="sitemap_index.news">News</a></nav><section class="sm-section" id="main"><header><h2><span data-i18n="sitemap_index.section_main">Main</span><span class="sm-count" data-count>62</span></h2>- <p class="sm-section-desc">Entry points into the product.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.entry_points_into_the_product">Entry points into the product.</p></header><ul class="sm-list"><li>@@ -649,7 +649,7 @@<section class="sm-section" id="build"><header><h2><span data-i18n="sitemap_index.section_build">Build</span><span class="sm-count" data-count>80</span></h2>- <p class="sm-section-desc">Create and customize an agent.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.create_and_customize_an_agent">Create and customize an agent.</p></header><ul class="sm-list"><li>@@ -1218,7 +1218,7 @@<section class="sm-section" id="labs"><header><h2><span data-i18n="sitemap_index.section_labs">Labs</span><span class="sm-count" data-count>25</span></h2>- <p class="sm-section-desc">Experimental features actively shipping. Expect fast iteration.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.experimental_features_actively_shipping_expe">Experimental features actively shipping. Expect fast iteration.</p></header><ul class="sm-list"><li>@@ -1401,8 +1401,8 @@<section class="sm-section" id="crypto"><header>- <h2><span data-i18n="sitemap_index.section_crypto">Crypto</span><span class="sm-count" data-count>152</span></h2>- <p class="sm-section-desc">On-chain features: pay-per-call, token launches, vanity wallets.</p>+ <h2><span data-i18n="sitemap_index.section_crypto">Crypto</span><span class="sm-count" data-count>153</span></h2>+ <p class="sm-section-desc" data-i18n="sitemap_index.on_chain_features_pay_per_call_token">On-chain features: pay-per-call, token launches, vanity wallets.</p></header><ul class="sm-list"><li>@@ -1678,6 +1678,13 @@<span class="sm-desc" data-i18n="sitemap_index.d_dashboard_capabilities">Live command center for all 4 autonomous agent capabilities: Alpha Hunt smart-money scoring, autonomous Coin Launcher on schedule, Creator Auto-Claim for fee harvesting, and Market Maker with Jito execution.</span></a></li>+ <li>+ <a href="/duels">+ <span class="sm-title"><span data-i18n="sitemap_index.t_duels">Trader Duels: call the winner</span></span>+ <span class="sm-path" translate="no">/duels</span>+ <span class="sm-desc" data-i18n="sitemap_index.d_duels">Call which of two pump.fun trader agents books more realized profit over a fixed UTC window. Free to play with a daily points allowance; points have no cash value. Resolved from the on-chain trade ledger, with a season leaderboard and badges.</span>+ </a>+ </li><li><a href="/arena"><span class="sm-title"><span data-i18n="sitemap_index.t_arena">The Arena: live PvP trading tournaments</span></span>@@ -2475,7 +2482,7 @@<section class="sm-section" id="agent-tools"><header><h2><span data-i18n="sitemap_index.section_agent-tools">Agent Tools</span><span class="sm-count" data-count>9</span></h2>- <p class="sm-section-desc">Manage, talk to, and deploy individual agents.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.manage_talk_to_and_deploy_individual_agents">Manage, talk to, and deploy individual agents.</p></header><ul class="sm-list"><li>@@ -2547,7 +2554,7 @@<section class="sm-section" id="account"><header><h2><span data-i18n="sitemap_index.section_account">Account</span><span class="sm-count" data-count>14</span></h2>- <p class="sm-section-desc">Sign-in, dashboard, billing.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.sign_in_dashboard_billing">Sign-in, dashboard, billing.</p></header><ul class="sm-list"><li>@@ -2654,7 +2661,7 @@<section class="sm-section" id="learn"><header><h2><span data-i18n="sitemap_index.section_learn">Learn</span><span class="sm-count" data-count>447</span></h2>- <p class="sm-section-desc">Docs, tutorials, and reference.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.docs_tutorials_and_reference">Docs, tutorials, and reference.</p></header><ul class="sm-list"><li>@@ -5792,7 +5799,7 @@<section class="sm-section" id="blog"><header><h2><span data-i18n="sitemap_index.section_blog">Blog</span><span class="sm-count" data-count>39</span></h2>- <p class="sm-section-desc">Long-form announcements, launches, and writing from three.ws.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.long_form_announcements_launches_and_writing">Long-form announcements, launches, and writing from three.ws.</p></header><ul class="sm-list"><li>@@ -6146,7 +6153,7 @@<section class="sm-section" id="machine"><header><h2><span data-i18n="sitemap_index.section_machine">Machine-readable</span><span class="sm-count" data-count>17</span></h2>- <p class="sm-section-desc">Indexes and well-known files for crawlers and AI agents.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.indexes_and_well_known_files_for_crawlers">Indexes and well-known files for crawlers and AI agents.</p></header><ul class="sm-list"><li>@@ -6274,7 +6281,7 @@<section class="sm-section" id="news"><header><h2><span data-i18n="sitemap_index.section_news">News</span><span class="sm-count" data-count>112</span></h2>- <p class="sm-section-desc">Product launches, integrations, and announcements from three.ws.</p>+ <p class="sm-section-desc" data-i18n="sitemap_index.product_launches_integrations_and_announceme">Product launches, integrations, and announcements from three.ws.</p></header><ul class="sm-list"><li>@@ -7065,10 +7072,10 @@</section><div class="sm-empty" id="sm-empty" hidden><p><span data-i18n="sitemap_index.no_pages_match">No pages match</span> “<span id="sm-empty-q"></span>”.</p>- <p class="sm-empty-hint">Try a shorter word, or search everything: global search covers agents, coins, and skills too.</p>+ <p class="sm-empty-hint" data-i18n="sitemap_index.try_a_shorter_word_or_search_everything">Try a shorter word, or search everything: global search covers agents, coins, and skills too.</p><div class="sm-empty-actions">- <button type="button" id="sm-empty-clear">Clear filter</button>- <a id="sm-empty-search" href="/search">Search all of three.ws</a>+ <button type="button" id="sm-empty-clear" data-i18n="sitemap_index.clear_filter">Clear filter</button>+ <a id="sm-empty-search" href="/search" data-i18n="sitemap_index.search_all_of_three_ws">Search all of three.ws</a></div></div></main>diff --git a/tests/api/llm-cooldown.test.js b/tests/api/llm-cooldown.test.jsnew file mode 100644index 000000000..1ea888cc5--- /dev/null+++ b/tests/api/llm-cooldown.test.js@@ -0,0 +1,154 @@+// The LLM chain's cross-request rung cooldown (api/_lib/llm.js). A rung that is+// out of credit, rejected, or retired gives the same answer for a long while, so+// later completions skip it instead of paying its round trip again; a 429 sits+// out only as long as the provider asks. It must never leave a request with+// nothing to try, and a transient 5xx must never cool a rung at all.++import { describe, it, expect, beforeEach, afterEach, vi } from 'vitest';+import {+ llmComplete,+ cooldownMsForStatus,+ llmCooldownSnapshot,+ resetLlmCooldowns,+} from '../../api/_lib/llm.js';++const KEYS = [+ 'GROQ_API_KEY',+ 'OPENROUTER_API_KEY',+ 'OPENROUTER_FALLBACK_KEYS',+ 'NVIDIA_API_KEY',+ 'NVIDIA_FALLBACK_KEYS',+ 'HF_TOKEN',+ 'HF_FALLBACK_TOKENS',+ 'LLM7_API_KEY',+ 'ANTHROPIC_API_KEY',+ 'OPENAI_API_KEY',+ 'GOOGLE_CLOUD_PROJECT',+];+const saved = {};++const ok = (content) => ({+ ok: true,+ status: 200,+ headers: new Headers(),+ json: async () => ({ choices: [{ message: { content } }], usage: { prompt_tokens: 1, completion_tokens: 1 } }),+ text: async () => '',+});+const fail = (status, headers = {}) => ({+ ok: false,+ status,+ headers: new Headers(headers),+ json: async () => ({}),+ text: async () => 'upstream said no',+});++// Only Groq (three rungs on one key) plus the two keyless lanes are in the+// chain, which keeps each case's fetch log short and readable.+function onlyGroq() {+ process.env.GROQ_API_KEY = 'g';+}++function routeFetch(handler) {+ const calls = [];+ globalThis.fetch = vi.fn(async (url, opts) => {+ const body = JSON.parse(opts.body);+ const hit = { host: new URL(String(url)).host, model: body.model };+ calls.push(hit);+ return handler(hit);+ });+ return calls;+}++beforeEach(() => {+ for (const k of KEYS) {+ saved[k] = process.env[k];+ delete process.env[k];+ }+ resetLlmCooldowns();+});+afterEach(() => {+ for (const k of KEYS) {+ if (saved[k] === undefined) delete process.env[k];+ else process.env[k] = saved[k];+ }+ vi.restoreAllMocks();+ vi.useRealTimers();+});++describe('cooldownMsForStatus', () => {+ it('sits a dead rung out for 30 minutes: no credit, bad key, retired model', () => {+ for (const status of [401, 402, 403, 404, 410]) expect(cooldownMsForStatus(status)).toBe(30 * 60_000);+ });++ it('honors Retry-After on a 429, capped at five minutes, 30s when absent', () => {+ expect(cooldownMsForStatus(429, '12')).toBe(12_000);+ expect(cooldownMsForStatus(429, '86400')).toBe(5 * 60_000);+ expect(cooldownMsForStatus(429, null)).toBe(30_000);+ expect(cooldownMsForStatus(429, 'not-a-number')).toBe(30_000);+ });++ it('never cools on a transient server error', () => {+ for (const status of [500, 502, 503, 504]) expect(cooldownMsForStatus(status)).toBe(0);+ });+});++describe('llmComplete with cooldowns', () => {+ it('skips a rung that 402d on the previous call and goes straight to the next', async () => {+ onlyGroq();+ let calls = routeFetch(({ model }) => (model === 'qwen/qwen3.8-27b' ? fail(402) : ok('from 120b')));+ const first = await llmComplete({ system: 's', user: 'u' });+ expect(first.provider).toBe('groq#120b');+ expect(calls.map((c) => c.model)).toEqual(['qwen/qwen3.8-27b', 'openai/gpt-oss-120b']);++ calls = routeFetch(() => ok('from 120b again'));+ const second = await llmComplete({ system: 's', user: 'u' });+ expect(second.provider).toBe('groq#120b');+ // The out-of-credit rung was not asked again.+ expect(calls.map((c) => c.model)).toEqual(['openai/gpt-oss-120b']);+ expect(llmCooldownSnapshot()).toEqual([+ expect.objectContaining({ rung: 'groq|qwen/qwen3.8-27b', status: 402 }),+ ]);+ });++ it('lets a 429 rung back in once its Retry-After has passed', async () => {+ vi.useFakeTimers({ toFake: ['Date'] });+ onlyGroq();+ routeFetch(({ model }) => (model === 'qwen/qwen3.8-27b' ? fail(429, { 'retry-after': '10' }) : ok('fallback')));+ await llmComplete({ system: 's', user: 'u' });++ vi.setSystemTime(Date.now() + 11_000);+ const calls = routeFetch(() => ok('primary is back'));+ const out = await llmComplete({ system: 's', user: 'u' });+ expect(out.provider).toBe('groq');+ expect(calls[0].model).toBe('qwen/qwen3.8-27b');+ });++ it('does not cool a rung on a 5xx', async () => {+ onlyGroq();+ routeFetch(({ model }) => (model === 'qwen/qwen3.8-27b' ? fail(503) : ok('fallback')));+ await llmComplete({ system: 's', user: 'u' });+ expect(llmCooldownSnapshot()).toEqual([]);+ });++ it('still tries every rung when all of them are cooling, rather than failing unasked', async () => {+ onlyGroq();+ routeFetch(() => fail(402));+ await expect(llmComplete({ system: 's', user: 'u' })).rejects.toMatchObject({ status: 502 });+ expect(llmCooldownSnapshot().length).toBeGreaterThan(0);++ const calls = routeFetch(({ model }) => (model === 'openai-fast' ? ok('keyless lane answered') : fail(402)));+ const out = await llmComplete({ system: 's', user: 'u' });+ expect(out.text).toBe('keyless lane answered');+ expect(calls.length).toBeGreaterThan(1);+ });++ it('clears a rung from cooldown the moment it answers again', async () => {+ onlyGroq();+ routeFetch(() => fail(402));+ await expect(llmComplete({ system: 's', user: 'u' })).rejects.toBeTruthy();+ routeFetch(() => ok('recovered'));+ const out = await llmComplete({ system: 's', user: 'u' });+ expect(out.text).toBe('recovered');+ expect(llmCooldownSnapshot().map((c) => c.rung)).not.toContain(`${out.provider}|${out.model}`);+ });+});diff --git a/tests/api/llm-free-chain-reachability.test.js b/tests/api/llm-free-chain-reachability.test.jsindex 7279702e7..22bb30469 100644--- a/tests/api/llm-free-chain-reachability.test.js+++ b/tests/api/llm-free-chain-reachability.test.js@@ -28,6 +28,7 @@ const HOSTS = {mistral: 'api.mistral.ai',zai: 'api.z.ai',cloudflare: 'api.cloudflare.com',+ huggingface: 'router.huggingface.co',ovh: 'oai.endpoints.kepler.ai.cloud.ovh.net',gemini: 'generativelanguage.googleapis.com',vertex: 'aiplatform.googleapis.com',@@ -47,11 +48,20 @@ const FREE_CHAIN = [{ provider: 'groq#120b', host: HOSTS.groq, model: 'openai/gpt-oss-120b' },{ provider: 'cerebras', host: HOSTS.cerebras, model: 'llama-3.3-70b' },{ provider: 'openrouter', host: HOSTS.openrouter, model: 'google/gemma-4-31b-it:free' },- { provider: 'nvidia', host: HOSTS.nvidia, model: 'nvidia/nemotron-3-super-120b-a12b' },+ // A saturated free model 429s every key at once, so the lane also rotates+ // across alternate ':free' models, each on its own upstream pool.+ { provider: 'openrouter:nemotron-3-super-120b-a12b', host: HOSTS.openrouter, model: 'nvidia/nemotron-3-super-120b-a12b:free' },+ { provider: 'openrouter:qwen3.8-27b', host: HOSTS.openrouter, model: 'qwen/qwen3.8-27b:free' },+ // One NIM rung per nvapi key; same host and model, told apart by the key.+ { provider: 'nvidia', host: HOSTS.nvidia, model: 'nvidia/nemotron-3-super-120b-a12b', auth: 'nvapi-x' },+ { provider: 'nvidia#2', host: HOSTS.nvidia, model: 'nvidia/nemotron-3-super-120b-a12b', auth: 'nvapi-y' },{ provider: 'sambanova', host: HOSTS.sambanova, model: 'Meta-Llama-3.3-70B-Instruct' },{ provider: 'mistral', host: HOSTS.mistral, model: 'mistral-small-latest' },{ provider: 'zai', host: HOSTS.zai, model: 'glm-4.7-flash' },{ provider: 'cloudflare', host: HOSTS.cloudflare, model: '@cf/meta/llama-3.3-70b-instruct-fp8-fast' },+ // Hugging Face Inference Providers, one rung per token's monthly credit.+ { provider: 'huggingface', host: HOSTS.huggingface, model: 'meta-llama/Llama-3.3-70B-Instruct', auth: 'hf-a' },+ { provider: 'huggingface#2', host: HOSTS.huggingface, model: 'meta-llama/Llama-3.3-70B-Instruct', auth: 'hf-b' },{ provider: 'ovh', host: HOSTS.ovh, model: 'Meta-Llama-3_3-70B-Instruct' },{ provider: 'gemini', host: HOSTS.gemini, model: 'gemini-2.5-flash-lite' },{ provider: 'vertex-gemini', host: HOSTS.vertex, model: 'google/gemini-2.5-flash' },@@ -67,6 +77,9 @@ const ENV_KEYS = ['OPENROUTER_API_KEY','OPENROUTER_FALLBACK_KEYS','NVIDIA_API_KEY',+ 'NVIDIA_FALLBACK_KEYS',+ 'HF_TOKEN',+ 'HF_FALLBACK_TOKENS','SAMBANOVA_API_KEY','MISTRAL_API_KEY','ZAI_API_KEY',@@ -95,6 +108,9 @@ function configureFreeLanes() {process.env.CEREBRAS_API_KEY = 'c';process.env.OPENROUTER_API_KEY = 'or';process.env.NVIDIA_API_KEY = 'nvapi-x';+ process.env.NVIDIA_FALLBACK_KEYS = 'nvapi-y';+ process.env.HF_TOKEN = 'hf-a';+ process.env.HF_FALLBACK_TOKENS = 'hf-b';process.env.SAMBANOVA_API_KEY = 'sn';process.env.MISTRAL_API_KEY = 'mi';process.env.ZAI_API_KEY = 'z';@@ -162,7 +178,11 @@ describe('free chain: every rung is reachable through a transport-level failure'// lane and the instant lane indistinguishable, which is exactly the// pair this suite has to tell apart.const requested = JSON.parse(opts.body).model;- const idx = FREE_CHAIN.findIndex((r) => u.includes(r.host) && r.model === requested);+ // Multi-key rungs share host AND model, so the key tells them apart.+ const auth = opts.headers?.authorization;+ const idx = FREE_CHAIN.findIndex(+ (r) => u.includes(r.host) && r.model === requested && (!r.auth || auth === `Bearer ${r.auth}`),+ );expect(idx, `unexpected fetch: ${u} (${requested})`).toBeGreaterThanOrEqual(0);tried.push(FREE_CHAIN[idx].provider);// A rung above the target dies at the transport level. Alternate the@@ -172,8 +192,11 @@ describe('free chain: every rung is reachable through a transport-level failure'// host, which is deliberate (three Groq rungs must not each burn the// budget while api.groq.com is hanging) and would make this case// measure the stall guard instead of chain reachability.- const sameHost = FREE_CHAIN[idx].host === rung.host;- if (idx < i) throw transportFailure(!sameHost && idx % 2 === 1 ? 'abort' : 'reset');+ // Same reasoning for any host that serves more than one rung (the+ // OpenRouter models, the NIM and Hugging Face keys): aborting one+ // would skip its siblings as a stalled host.+ const sharedHost = FREE_CHAIN.filter((r) => r.host === FREE_CHAIN[idx].host).length > 1;+ if (idx < i) throw transportFailure(!sharedHost && idx % 2 === 1 ? 'abort' : 'reset');return okOpenAiShape(`served by ${FREE_CHAIN[idx].provider}`, rung.model);});diff --git a/tests/api/llm.test.js b/tests/api/llm.test.jsindex e754e3a55..47a1a1864 100644--- a/tests/api/llm.test.js+++ b/tests/api/llm.test.js@@ -209,19 +209,17 @@ describe('llmComplete — multiple OpenRouter keys', () => {it('dedupes a fallback key that repeats the primary', async () => {process.env.OPENROUTER_API_KEY = 'or-same';process.env.OPENROUTER_FALLBACK_KEYS = 'or-same, or-extra';- let n = 0;- globalThis.fetch = vi.fn(async () => {- n += 1;+ const primaryModelKeys = [];+ globalThis.fetch = vi.fn(async (url, opts) => {+ if (String(url).includes('openrouter.ai') && JSON.parse(opts.body).model === 'google/gemma-4-31b-it:free') {+ primaryModelKeys.push(opts.headers.authorization);+ }return errResp(500);});await expect(llm.llmComplete({ system: 's', user: 'u' })).rejects.toMatchObject({ status: 502 });- // or-same is deduped to a single key (one :free rung), then or-extra's :free- // rung, two OpenRouter fetches (without dedup or-same would be tried again as- // a fallback too). The chain then falls through the two unconditional keyless- // lanes (OVH, Pollinations) before giving up: four fetches total. LLM7 is not- // among them any more; llm7.io retired its anonymous tier, so that rung is- // gated on LLM7_API_KEY (see api/_lib/llm.js).- expect(n).toBe(4);+ // or-same is deduped to a single key, so the primary free model is asked+ // exactly once per distinct key (without dedup or-same would be tried twice).+ expect(primaryModelKeys).toEqual(['Bearer or-same', 'Bearer or-extra']);});it('llmConfigured is true with only fallback keys set', () => {diff --git a/tests/setup.llm-cooldowns.js b/tests/setup.llm-cooldowns.jsnew file mode 100644index 000000000..9a8cfbda3--- /dev/null+++ b/tests/setup.llm-cooldowns.js@@ -0,0 +1,11 @@+// The LLM chain remembers rungs that recently failed (api/_lib/llm.js, provider+// cooldown) for the life of the server instance. In tests that memory would+// leak between cases: a rung one test fails with a 429 would be skipped by the+// next test that expects it to answer. Clear it before every test. The map+// lives on a global symbol precisely so this file never imports llm.js, which+// would load it ahead of each test file's own vi.mock() calls.+import { beforeEach } from 'vitest';++beforeEach(() => {+ globalThis[Symbol.for('three.ws.llmCooldowns')]?.clear();+});diff --git a/vitest.config.js b/vitest.config.jsindex 33780b4a6..147582a50 100644--- a/vitest.config.js+++ b/vitest.config.js@@ -64,7 +64,7 @@ export default defineConfig({// enough to decide the home lane's local-instance seam: the SSRF guard// reads it once at its own import and never again, on purpose. Inert// unless a live Home Assistant run was asked for. See the file.- setupFiles: ['./tests/setup.home-seam.js'],+ setupFiles: ['./tests/setup.home-seam.js', './tests/setup.llm-cooldowns.js'],include: ['tests/**/*.test.js','tests/**/*.test.mjs',