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): publish the concierge, walk leaderboard, and embedded-voice entries
nirholas9d6e7aaf7a18
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 3bff623b2..95c890f46 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -19,6 +19,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr- **Agent .sol name lookups work again, and the avatar manifest schema now answers at its own address**: The avatar-agent MCP server resolves .eth and .sol names to wallet addresses. Its .sol side went through a hosted name-lookup service that has since been retired: every request came back as "not found", which reads exactly like an unregistered domain, so a working name and a nonexistent one were indistinguishable. Solana names are now read straight from the chain, from the same name-service accounts the three.ws backend already reads, so the lookup depends on nothing but a Solana RPC. Two things were fixed alongside it. The per-lookup time budget was under five seconds, which a cold Ethereum lookup routinely exceeds, so real names were being reported as timeouts; it is now fifteen seconds and adjustable. And a lookup that genuinely fails now says why instead of quietly reporting "not found". Separately, the avatar manifest schema names https://three.ws/schema/avatar.v1.json as its own identifier, but nothing was served there, so any validator that followed the identifier got a 404. The schema is now published at that address, with a test that keeps the served copy identical to the one shipped in the package. `[fix, sdk]`- **Agent coin pages stop reporting a burn that never happened**: An agent coin page showed a "last burn" time taken from the last moment the buyback job ran at all, including every run where it looked and correctly decided there was nothing to burn. A coin with thousands of those skipped runs and no actual burns therefore read as having burned minutes ago while the burn count and burned total beside it both sat at zero. The burn time now counts confirmed burns only, so it stays blank until a real burn lands and always agrees with the count next to it. Separately, the three pump endpoints that manage a coin's autopilot policy and its invoices now require a request to originate from three.ws before it can act on a signed-in browser session, which is the rule the rest of the pump endpoints already followed. Agent keys and workers are unaffected. `[fix, security]`- **Agent Screen is usable on a phone, and reads correctly to a screen reader**: Watching an agent work on a narrow screen was close to impossible. The row of viewer controls along the top (fit, screenshot, picture-in-picture, fullscreen, ambient, zen, and the link back to the agent) ran off the edge of the page with no way to reach them, the agent name and its live badge painted over the buttons underneath, and the two panels that open by default covered the live screen and the avatar cam completely: the whole point of the page, hidden. The control row now scrolls sideways so every control stays reachable at any width, the name keeps its space, and on a phone the page opens on a bare stage with the panels one tap away. Three fixes for anyone using a keyboard or a screen reader landed with it: the page now has a real heading and a main region for the skip link to jump to, the hidden Skip to content link stopped rendering as visible text at the top of the page, and every text field on the page gained a proper label. Finally, if the agent list fails to load for any reason other than being signed out, the page says what went wrong and offers a retry instead of telling a signed-in owner to go sign in. (`/agent-screen`) `[fix, improvement]`+- **An embedded agent keeps talking when the browser refuses to play its voice**: The living agent body that renders inside ChatGPT, Claude, and any site that embeds it speaks each reply with a lip-synced mouth. When a voice track is attached, the body plays the audio and drives the mouth from it. Browsers block that playback routinely: a chat panel the visitor has not clicked in is not allowed to start audio. The body was falling back to an empty line in that case, so it stood there with its mouth shut through the whole reply. It now falls back to the text of the reply it was actually given, so the mouth speaks the line whether or not the audio was allowed to start, and the engine is now covered by tests for the load, speak, emotion, chain-state, and teardown paths. `[fix]`- **Arming an Oracle agent for live trading no longer rounds your trade size up**: The arm page clamps the risk knobs on the server so a bad number can never reach the trading loop, and those clamps have a floor: the smallest trade the loop will place is 0.001 SOL. That floor was being applied in the wrong direction. Ask to arm a live agent with a per-trade size of zero, and instead of being told no, you were armed for real spending at 0.001 SOL a trade with a half-SOL daily ceiling you never asked for. The check meant to catch this looked at the number after the clamp had already raised it, so it could never fire. Arming live now refuses any per-trade size below the floor, and any daily cap smaller than a single trade, and tells you which number to fix; nothing is armed until you do. Clamping a too-large number back down to the ceiling is unchanged, and simulate runs keep the same forgiving behaviour as before. The arm page also shows the reason a save was rejected instead of one generic line, so an ownership problem no longer reads the same as a bad trade size, and a failed Telegram test now prints what Telegram actually said. (`/oracle/arm`) `[fix, security]`- **Asking for a voice that does not exist now tells you so, instead of quietly speaking in a different one**: The voice API accepts a provider and a voice id from the voice catalog. If the id was not one that provider actually has, nothing said so: the request was answered with audio rendered in that lane's default voice, and on the metered lanes it was billed, so a single typo in a voice id came back as a stranger's voice at full price with a success status. A well-formed id for a voice that had been removed took the opposite route and failed as a server error, which reads as our fault and tells a client to retry something that will never work. Both now return a plain rejection that names the id you sent and the catalog address to pick a real one from, and that check runs before anything is charged. Leaving the voice id out still gets you the lane default on purpose. Asking the catalog for a provider that does not exist used to return an empty list that was indistinguishable from a provider being down; it now says which providers exist. ElevenLabs voices, whose catalog grows every time you clone or add one, are still decided by ElevenLabs, but an id it rejects is reported as your request being wrong rather than ours, with any charge refunded first. Voice cloning also now asks who you are before reporting anything about how this server is configured. In the same pass, a key ElevenLabs turns down when you list your voices or browse the public voice library now comes back as your key being rejected, so the page can ask you for a new one, instead of a gateway error that invites a retry which can never succeed. (`/docs/api-reference`) `[fix, improvement]`- **Avatars can move their lips again: the Audio2Face lane is back**: Every call that turns speech into facial animation had been failing with an upstream not-found error, so a talking avatar spoke with a frozen face and the one-shot say() path returned nothing usable. The hosted model id the lane pointed at had stopped resolving for our NVIDIA account. It now points at the hosted Audio2Face NIM that is actually live, verified end to end: a synthesized clip drives a 30 fps ARKit blendshape track whose jaw genuinely opens and closes with the words. The discovery helper that finds these model ids was missing the name NVIDIA publishes them under, which is why the outage looked like a configuration mystery; it now finds them. `[fix]`@@ -52,6 +53,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr- **The Agent Exchange stops offering a purchase its wallet cannot pay for**: On /agent-exchange two 3D agents buy and sell live crypto intel, and the buyer settles $0.01 of real USDC on Solana for every signal. The page already warned you when the buyer's wallet was too empty to settle, but it left the Buy intel button live anyway, so the only thing a click could produce was a failed payment. The button now turns itself off and says why, and the warning carries a Check again button, so a wallet you fund while the page is open recovers without a reload. The balance is also re-read after every purchase, so a settlement that empties the wallet is reflected immediately instead of one click too late. When the balance check itself does not answer, the page used to say nothing at all and quietly log a browser error; it now tells you the balance is unknown and offers the same retry. Two smaller fixes came with it: the page title and the buy button were being rewritten at load time from the translation catalog, which had drifted from the page and still carried the old punctuation, now corrected in all 85 languages; and the two avatar frames were keyboard stops that did nothing and showed no focus outline, so they are out of the tab order and the topic chips and buy button are two tabs closer for anyone navigating by keyboard. (`/agent-exchange`) `[fix, improvement]`- **The agent-payment and MCP docs now show the code that actually ships**: Three developer docs had drifted from the code they document, in ways that cost a reader real time. The Payment Sessions walkthrough, which is the guide for giving an agent a budget instead of a wallet key, quoted five snippets that no longer match the source: the branch that refuses a payment when the budget will not cover it, the host-allowlist check (which has moved into a shared module so the dry-run simulator and the live enforcement can never disagree), the expiry sweep that refunds unspent budget, the dollars-to-atomic-units helper, and a column missing from the table definition. All five now match the shipped code line for line, and the page links the two source files it had not been pointing at. The MCP article listed the Solana vanity grinder at up to $0.50 scaled by the work done; it is a flat $0.05 per successful grind, and its setup example set a wallet variable the package does not read, so anyone copying it would have configured a payout address that was silently ignored. That example now shows only the variable that works, and says plainly that settlement is USDC on Solana mainnet. The IBM Granite forecasting example showed three data points for a tool that requires at least sixty-four, so a copied sample was guaranteed to be rejected; it now says so. (`/docs/aws-builder-center-agent-payment-sessions`) `[docs, fix]`- **The Alibaba Cloud MCP server starts now, and five MCP packages document how to import them**: The @three-ws/alibaba-cloud-mcp package, which lets an agent use Qwen chat, embeddings, and model discovery on your own Alibaba Cloud key, could not start at all. Two faults, both on the first line of use: the check for whether the file was launched directly crashed before the server ever opened, and the three tools were registered with a schema format the MCP library rejects. Both are fixed, so the server now boots, advertises its tools, and answers a client. The payment-sessions MCP package had a documentation fault with the same cost: its quick start told you to set your session variable to the whole cookie including the name, which the server then wrapped in the cookie name again, so every session-management call came back unauthorized. It now says to use the cookie value, and ships its two runnable examples inside the published package instead of only in the repository. Five packages (payment sessions, Agora, alerts, the assistant widget, and Alibaba Cloud) gained a Programmatic use section showing the exact code to import their tools into a host of your own, and the sniper package's HTTP table now lists every endpoint it actually serves, including the paid close-position route it had left out. `[fix, sdk, docs]`+- **The concierge widget closes on Escape from anywhere on the page, and its SDK now installs on a server**: Three fixes to the embeddable concierge. Escape now closes the chat panel wherever the visitor's cursor last landed: it only worked while focus was still inside the widget, so clicking the page behind the panel left the documented shortcut dead. Sites built with Next, Nuxt, SvelteKit, or Astro can now import the package the normal way, because importing it no longer expects a browser and no longer crashes the server render. And the 3D avatar stage stopped printing a deprecation warning into every host site's console. The package also ships full TypeScript types for shopping mode (catalog, products, policies, and the store events) and declares its own build and test tools, so a standalone install builds and tests without the monorepo. (`/concierge`) `[fix, sdk]`- **The contracts README now describes the contracts that are actually there**: Anyone opening the on-chain contracts directory was reading a page written when it held three ERC-8004 registries. It now holds ten contracts and programs, including the two that custody real USDC on BNB Smart Chain, Base and Arbitrum, and the two Solana programs. The README listed none of them, claimed 25 tests where there are 242, and printed the canonical ERC-8004 identity addresses with mainnet and testnet swapped, so a developer copying them would have pointed at the wrong chain's registry. It also told you to edit a shared entry that seven testnets read from, which would have silently repointed all of them. All of that is corrected, the Solana build and test commands are documented, and the sample registration snippet now shows the values the function really returns. `[docs]`- **The Duplicate button on your widgets works again**: Duplicating a widget from the dashboard failed every single time. The clone request never gave the new widget an identifier, and because the database requires one, the copy was rejected before it was written and the page showed only a generic server error with a support reference. A second, correct copy of the same clone logic was already in the codebase and the route simply was not using it; the route now runs that one, so duplicating gives you a real new widget with its own id, its own name ending in (copy), and the original's type, look, avatar and visibility carried over. The duplicate request also now carries the same cross-site request protection every other widget change already had, and it accepts an API key with write access, so you can clone a widget from a script as well as from the dashboard. Two smaller corrections travelled with it. When no chat provider is configured and the platform answers a visitor with its own placeholder line, that message is no longer filed in your transcript as though a model had written it. And chatting with one of the showcase widgets in the public gallery no longer tries to file a transcript against a widget that does not exist, which failed on every message. The developer reference now documents duplicate, stats, chat, transcripts and the knowledge base, all of which were live but undocumented. (`/dashboard/widgets`) `[fix, docs]`- **The embed and sharing tutorials now describe the surfaces that actually exist**: An audit of eight beginner tutorials found several describing features the platform does not have. The avatar-swap tutorial sent readers to Widget Studio for a body swap it cannot do, through a gallery with filters, preview buttons, and a 'Bust cache' save dialog that were never built; it now documents the real one-click swap in the agent editor's Outfit panel, and explains why nothing needs a cache purge. The sharing tutorial claimed rich social previews came from the plain agent URL, which actually serves a generic card, and described a headless renderer, a 24-hour image cache, a 'Refresh preview image' button, and a private-profile toggle that do not exist; it now points at the real share URL that carries the agent's name, description, and card, and explains what genuinely controls discoverability. The AR tutorial's iframe example used an address that returns 404 for an avatar id. The embed quickstart showed an agent ID in the wrong format, overstated the bundle size, and had the effect of the shadow DOM backwards. The docs-world tutorial gained the search palette it never mentioned, and the greeting tutorial now warns that the editor's Greeting field is not what makes an embed talk. Five page descriptions were corrected too, including one promising knobs the tutorial never covered. (`/tutorials/swap-avatar-in-studio`) `[docs, fix]`@@ -74,6 +76,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr- **The Tutor’s running tab and end-of-session invoice now show what you actually paid**: The Tutor charges one cent per answer and keeps an itemized tab you can reload later and close for a signed invoice. That ledger only ever knew how to store itself in a key-value service this deployment does not run, so nothing was ever written down: reopening the page never restored your history, and ending a session always produced an invoice reading "no questions" and a total of zero, however many answers you had paid for. The tab is now stored in the platform’s own database, so it survives a reload, a new browser tab, and any server restart, and the invoice lists every answer with its charge and a verification hash over the line items. Closing a session twice returns the identical invoice instead of a freshly stamped one, closing a session id that was never used no longer makes that id unusable for its real owner, and if the ledger itself is ever unreachable the page says so rather than quietly reporting that you owe nothing. Asking a question still succeeds even if the ledger write fails, because a bookkeeping outage should never cost you the answer you paid for. (`/tutor`) `[fix]`- **The vanity receipt verifier stops calling honest receipts fraudulent, and the avatar CLI takes a Solana address**: Two documented commands did not do what their page said. The command-line verifier for a provably-fair vanity address receipt scored the difficulty claim with the retired flat model, the one that treats every Base58 character as equally likely. Every receipt issued since the leading-character correction landed therefore failed its difficulty check and was reported as untrustworthy, even though the in-browser verifier at /vanity/verify passed the same receipt. The verifier now reads the model each receipt names in its own signed difficulty block, exactly as the page and the SDK do, so current receipts verify and older ones keep verifying under the model they were issued under. Separately, the Rig Doctor page showed an avatar-manifest command taking a Solana address as the owner, which the CLI rejected outright: it understood Ethereum addresses and full CAIP-10 identifiers and nothing else. Since Solana is the home chain, a bare Solana address is now accepted and recorded as solana:mainnet-beta. Alongside those, eight tutorial and reference pages were audited against the code they describe: the Vercel self-hosting guide had a stale copy of the config file and pointed at a migrations folder that has moved, the MCP guide listed three placeholder OAuth scopes where the live server publishes twelve real per-capability ones, and the Avatar CLI page pinned a version number that npm has never had. (`/docs/rig-doctor`) `[fix, docs, sdk]`- **The vanity SDK now uses the payment wallet you hand it, and the in-chat 3D avatar keeps its tuned lighting**: Three fixes in the developer packages. The @three-ws/vanity SDK documented a fetch option on its paid address-grinding lane but silently ignored it, so a developer who passed a payment-enabled fetch got a payment-required error back instead of the keypair they paid for; that option now works, as do a custom API host and key. The avatar MCP server's in-chat 3D viewer had been hand-tuned with softer shadows, brighter exposure and a neutral environment light, but the script that rebuilds it before every release still produced the old flat version, so the next release would have quietly undone the tuning; the script now produces the tuned viewer, and the same settings apply to the fallback view for chat apps that cannot run the interactive one. The tool SDK's quick-start example pointed at an API that does not exist and skipped the wrapper an MCP server needs to return anything at all, so following it produced a tool that answered with nothing; it now runs end to end against a real free three.ws endpoint. `[fix, sdk, docs]`+- **The Walk page shows the real top walkers again instead of claiming nobody has walked**: The Top walkers board on /walk read "No walks logged yet. Be the first to put down some distance" for every visitor, while twenty-nine people had in fact logged distance and the full leaderboard at /walk-leaderboard listed all of them. The page was asking the leaderboard endpoint for the ranking and getting a correct answer back, then looking for it under half a dozen field names the endpoint has never used, finding nothing, and drawing its empty state. It now reads the shape the endpoint actually returns, so the top three walkers appear with their handles, avatars, distances, and a link to each profile, and it asks for only the three rows it draws instead of the first fifty. The social section under it now shows placeholder cards while it loads rather than a blank gap. (`/walk`) `[fix]`- **Three docs pages stop pointing readers at a dead page, a missing section, and a snippet that does nothing**: The Vanity page told you to grind a long address yourself in the browser at /vanity, but that path has never existed and answers 404; the in-browser grinder lives at /vanity-wallet, and both links on the page now go there. In Troubleshooting, the fix for a Quick Look model that opens and instantly closes linked to a model-optimization section that is not on the AR page, so the link dropped you at the top with nothing to read; it now opens the section that actually covers it. The same page told you to add a responsive attribute to size an inline embed, with a snippet that set the width in CSS. Neither half worked: responsive sizing is already on by default, and the element reads the width attribute rather than the computed style, so the documented snippet never produced the 3:4 ratio it promised. The instructions now show the form that does work, and name the two things that switch it off. (`/docs/troubleshooting`) `[fix, docs]`- **Two tutorials stop shipping code samples that quietly do nothing**: An audit of eight developer tutorials found two whose main code sample looked right and silently failed. The custom-skill weather handler split the weather API's reply on a plus sign, but the temperature field arrives with a leading plus of its own, so every field after the condition landed one slot off and the temperature always came back empty; it also mapped the fahrenheit option to a format code that means feels-like rather than Fahrenheit, so the celsius and fahrenheit paths returned the same units. Both are fixed and the sample was run against the live API in both unit modes. The gesture tutorial's AI lane read an emotion-breakdown field the Granite analysis tool has never returned, so the strength check never passed and the agent never played a single animation; it now reads the score and risk flags the tool actually returns, and grades the gesture across four bands instead of a binary. Separately, twelve tutorial hero buttons pointed at anchors on the docs hub that stopped existing when the hub split into per-topic pages, and now open the real pages; a placeholder URL in the Pump.fun tutorial was being parsed as HTML, which broke a link and injected stray elements into the page; and the self-host guide was corrected throughout, including storage variable names that no longer matched the code, a database step that left most tables uncreated, an audit-logging example importing a function that does not exist, a Cloudflare worker setup for a proxy the platform does not run, and a license claim that described the repository as open source when it is not. (`/tutorials/custom-skill`) `[docs, fix]`- **Voice and vision SDKs ship runnable examples, and the x402 seller SDK reads its Solana sponsor from the environment**: @three-ws/voice and @three-ws/vision-mcp now include example scripts you can run as-is against the live free lanes: the voice one speaks a sentence, animates a face from those exact bytes, and transcribes it back; the vision one probes the lane, writes alt text for an image, and reads the text off it. @three-ws/x402-server used to demand the Solana sponsor account as a code option even though its own error message told you to set X402_FEE_PAYER_SOLANA, so a correctly configured deployment still got missing_fee_payer; it now reads that environment variable when the option is omitted. The VS Code x402 extension's endpoint scaffolder was generating Base-only endpoints and now generates Solana-first ones, matching every paid route on the platform, and it ships with a test suite covering which payment requirement it selects. `[sdk, fix]`@@ -3018,7 +3021,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr- **Pole Club** (`/club`): A 3D club where dancers only perform when you send a micro-tip. Pay $0.001 per routine, settled on-chain via x402 on Solana.- **Scan: Selfie to 3D Avatar** (`/features/scan`): Point your camera at your face, hold still, and walk away with a rigged 3D avatar in 60 seconds. Free, in your browser.- **Studio: Embeddable AI Widget Builder** (`/features/studio`): Configure avatar, voice, and knowledge base in the Widget Studio, then copy one script tag to embed a 3D AI agent anywhere.-- **Voice Lab** (`/voice`): Browse about 850 voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.+- **Voice Lab** (`/voice`): Browse hundreds of voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.- **Walk: 3D Avatar AR** (`/features/walk`): Drive your AI agent's 3D avatar with WASD or joystick. Toggle AR mode and it walks on your real floor through your phone camera.- **What is three.ws?** (`/what-is`): Plain-English introduction to three.ws: what a 3D AI agent is, real use-cases, and where to begin.- **Cosmetics economy in the game world**: A cosmetics shop inside the 3D game world, buy avatar customizations via x402, trade them peer-to-peer, and earn from cosmetics usage. `[feature]`diff --git a/data/changelog.json b/data/changelog.jsonindex 4573e6fe3..7d099204b 100644--- a/data/changelog.json+++ b/data/changelog.json@@ -1,6 +1,43 @@{"$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": "2026-08-15",+ "title": "The multiplayer worlds server connects on any local port, and its README documents what it actually reads",+ "summary": "Two fixes to the server that runs every live world (/play, /walk, /agora, /irl, /play/war, the Living Stages, /ar/studio). Running two copies of the site on one machine used to break the worlds: the second copy gets the next free web port, and the server only accepted connections from the first four, so the world silently refused to join. Outside production it now accepts any connection coming from the same machine, while the live server keeps its exact origin list unchanged. Its README also described six settings when the server reads more than thirty, and listed two web endpoints when it serves seven, so anyone running or deploying it was working from a map that no longer matched the ground; the settings, the endpoints, and how the admin monitor is protected are now written down as they really behave, and the origin rules, the refusal to boot production without a real signing secret, and the closed admin monitor are covered by tests.",+ "tags": [+ "fix",+ "docs",+ "infra"+ ]+ },+ {+ "date": "2026-08-15",+ "title": "An embedded agent keeps talking when the browser refuses to play its voice",+ "summary": "The living agent body that renders inside ChatGPT, Claude, and any site that embeds it speaks each reply with a lip-synced mouth. When a voice track is attached, the body plays the audio and drives the mouth from it. Browsers block that playback routinely: a chat panel the visitor has not clicked in is not allowed to start audio. The body was falling back to an empty line in that case, so it stood there with its mouth shut through the whole reply. It now falls back to the text of the reply it was actually given, so the mouth speaks the line whether or not the audio was allowed to start, and the engine is now covered by tests for the load, speak, emotion, chain-state, and teardown paths.",+ "tags": [+ "fix"+ ]+ },+ {+ "date": "2026-08-15",+ "title": "The concierge widget closes on Escape from anywhere on the page, and its SDK now installs on a server",+ "summary": "Three fixes to the embeddable concierge. Escape now closes the chat panel wherever the visitor's cursor last landed: it only worked while focus was still inside the widget, so clicking the page behind the panel left the documented shortcut dead. Sites built with Next, Nuxt, SvelteKit, or Astro can now import the package the normal way, because importing it no longer expects a browser and no longer crashes the server render. And the 3D avatar stage stopped printing a deprecation warning into every host site's console. The package also ships full TypeScript types for shopping mode (catalog, products, policies, and the store events) and declares its own build and test tools, so a standalone install builds and tests without the monorepo.",+ "tags": [+ "fix",+ "sdk"+ ],+ "link": "/concierge"+ },+ {+ "date": "2026-08-15",+ "title": "The Walk page shows the real top walkers again instead of claiming nobody has walked",+ "summary": "The Top walkers board on /walk read \"No walks logged yet. Be the first to put down some distance\" for every visitor, while twenty-nine people had in fact logged distance and the full leaderboard at /walk-leaderboard listed all of them. The page was asking the leaderboard endpoint for the ranking and getting a correct answer back, then looking for it under half a dozen field names the endpoint has never used, finding nothing, and drawing its empty state. It now reads the shape the endpoint actually returns, so the top three walkers appear with their handles, avatars, distances, and a link to each profile, and it asks for only the three rows it draws instead of the first fifty. The social section under it now shows placeholder cards while it loads rather than a blank gap.",+ "tags": [+ "fix"+ ],+ "link": "/walk"+ },{"date": "2026-08-15","title": "The trading control room now tells you when the agents cannot afford to trade",diff --git a/public/changelog.json b/public/changelog.jsonindex fdd5e97de..f4103d8b5 100644--- a/public/changelog.json+++ b/public/changelog.json@@ -1,5 +1,5 @@{- "generated_at": "2026-08-15T19:03:12.893Z",+ "generated_at": "2026-08-15T19:11:05.234Z","generated_by": "scripts/build-page-index.mjs from data/pages.json + data/changelog.json","site": {"name": "three.ws",@@ -146,6 +146,16 @@"improvement"]},+ {+ "date": "2026-08-15",+ "type": "update",+ "title": "An embedded agent keeps talking when the browser refuses to play its voice",+ "summary": "The living agent body that renders inside ChatGPT, Claude, and any site that embeds it speaks each reply with a lip-synced mouth. When a voice track is attached, the body plays the audio and drives the mouth from it. Browsers block that playback routinely: a chat panel the visitor has not clicked in is not allowed to start audio. The body was falling back to an empty line in that case, so it stood there with its mouth shut through the whole reply. It now falls back to the text of the reply it was actually given, so the mouth speaks the line whether or not the audio was allowed to start, and the engine is now covered by tests for the load, speak, emotion, chain-state, and teardown paths.",+ "link": null,+ "tags": [+ "fix"+ ]+ },{"date": "2026-08-15","type": "update",@@ -503,6 +513,17 @@"docs"]},+ {+ "date": "2026-08-15",+ "type": "update",+ "title": "The concierge widget closes on Escape from anywhere on the page, and its SDK now installs on a server",+ "summary": "Three fixes to the embeddable concierge. Escape now closes the chat panel wherever the visitor's cursor last landed: it only worked while focus was still inside the widget, so clicking the page behind the panel left the documented shortcut dead. Sites built with Next, Nuxt, SvelteKit, or Astro can now import the package the normal way, because importing it no longer expects a browser and no longer crashes the server render. And the 3D avatar stage stopped printing a deprecation warning into every host site's console. The package also ships full TypeScript types for shopping mode (catalog, products, policies, and the store events) and declares its own build and test tools, so a standalone install builds and tests without the monorepo.",+ "link": "/concierge",+ "tags": [+ "fix",+ "sdk"+ ]+ },{"date": "2026-08-15","type": "update",@@ -740,6 +761,16 @@"docs"]},+ {+ "date": "2026-08-15",+ "type": "update",+ "title": "The Walk page shows the real top walkers again instead of claiming nobody has walked",+ "summary": "The Top walkers board on /walk read \"No walks logged yet. Be the first to put down some distance\" for every visitor, while twenty-nine people had in fact logged distance and the full leaderboard at /walk-leaderboard listed all of them. The page was asking the leaderboard endpoint for the ranking and getting a correct answer back, then looking for it under half a dozen field names the endpoint has never used, finding nothing, and drawing its empty state. It now reads the shape the endpoint actually returns, so the top three walkers appear with their handles, avatars, distances, and a link to each profile, and it asks for only the three rows it draws instead of the first fifty. The social section under it now shows placeholder cards while it loads rather than a blank gap.",+ "link": "/walk",+ "tags": [+ "fix"+ ]+ },{"date": "2026-08-15","type": "update",@@ -29348,7 +29379,7 @@"date": "2026-06-05","type": "launch","title": "Voice Lab",- "summary": "Browse about 850 voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.",+ "summary": "Browse hundreds of voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.","link": "/voice","tags": ["launch"diff --git a/public/changelog.xml b/public/changelog.xmlindex 2180b94fe..763b30db9 100644--- a/public/changelog.xml+++ b/public/changelog.xml@@ -110,6 +110,14 @@<category>fix</category><description>Watching an agent work on a narrow screen was close to impossible. The row of viewer controls along the top (fit, screenshot, picture-in-picture, fullscreen, ambient, zen, and the link back to the agent) ran off the edge of the page with no way to reach them, the agent name and its live badge painted over the buttons underneath, and the two panels that open by default covered the live screen and the avatar cam completely: the whole point of the page, hidden. The control row now scrolls sideways so every control stays reachable at any width, the name keeps its space, and on a phone the page opens on a bare stage with the panels one tap away. Three fixes for anyone using a keyboard or a screen reader landed with it: the page now has a real heading and a main region for the skip link to jump to, the hidden Skip to content link stopped rendering as visible text at the top of the page, and every text field on the page gained a proper label. Finally, if the agent list fails to load for any reason other than being signed out, the page says what went wrong and offers a retry instead of telling a signed-in owner to go sign in.</description></item>+ <item>+ <title>An embedded agent keeps talking when the browser refuses to play its voice</title>+ <link>https://three.ws/changelog</link>+ <guid isPermaLink="false">2026-08-15:An embedded agent keeps talking when the browser refuses to play its voice</guid>+ <pubDate>Sat, 15 Aug 2026 12:00:00 GMT</pubDate>+ <category>fix</category>+ <description>The living agent body that renders inside ChatGPT, Claude, and any site that embeds it speaks each reply with a lip-synced mouth. When a voice track is attached, the body plays the audio and drives the mouth from it. Browsers block that playback routinely: a chat panel the visitor has not clicked in is not allowed to start audio. The body was falling back to an empty line in that case, so it stood there with its mouth shut through the whole reply. It now falls back to the text of the reply it was actually given, so the mouth speaks the line whether or not the audio was allowed to start, and the engine is now covered by tests for the load, speak, emotion, chain-state, and teardown paths.</description>+ </item><item><title>Arming an Oracle agent for live trading no longer rounds your trade size up</title><link>https://three.ws/oracle/arm</link>@@ -374,6 +382,14 @@<category>fix</category><description>The @three-ws/alibaba-cloud-mcp package, which lets an agent use Qwen chat, embeddings, and model discovery on your own Alibaba Cloud key, could not start at all. Two faults, both on the first line of use: the check for whether the file was launched directly crashed before the server ever opened, and the three tools were registered with a schema format the MCP library rejects. Both are fixed, so the server now boots, advertises its tools, and answers a client. The payment-sessions MCP package had a documentation fault with the same cost: its quick start told you to set your session variable to the whole cookie including the name, which the server then wrapped in the cookie name again, so every session-management call came back unauthorized. It now says to use the cookie value, and ships its two runnable examples inside the published package instead of only in the repository. Five packages (payment sessions, Agora, alerts, the assistant widget, and Alibaba Cloud) gained a Programmatic use section showing the exact code to import their tools into a host of your own, and the sniper package's HTTP table now lists every endpoint it actually serves, including the paid close-position route it had left out.</description></item>+ <item>+ <title>The concierge widget closes on Escape from anywhere on the page, and its SDK now installs on a server</title>+ <link>https://three.ws/concierge</link>+ <guid isPermaLink="false">2026-08-15:The concierge widget closes on Escape from anywhere on the page, and its SDK now installs on a server</guid>+ <pubDate>Sat, 15 Aug 2026 12:00:00 GMT</pubDate>+ <category>fix</category>+ <description>Three fixes to the embeddable concierge. Escape now closes the chat panel wherever the visitor's cursor last landed: it only worked while focus was still inside the widget, so clicking the page behind the panel left the documented shortcut dead. Sites built with Next, Nuxt, SvelteKit, or Astro can now import the package the normal way, because importing it no longer expects a browser and no longer crashes the server render. And the 3D avatar stage stopped printing a deprecation warning into every host site's console. The package also ships full TypeScript types for shopping mode (catalog, products, policies, and the store events) and declares its own build and test tools, so a standalone install builds and tests without the monorepo.</description>+ </item><item><title>The contracts README now describes the contracts that are actually there</title><link>https://three.ws/changelog</link>@@ -550,6 +566,14 @@<category>fix</category><description>Three fixes in the developer packages. The @three-ws/vanity SDK documented a fetch option on its paid address-grinding lane but silently ignored it, so a developer who passed a payment-enabled fetch got a payment-required error back instead of the keypair they paid for; that option now works, as do a custom API host and key. The avatar MCP server's in-chat 3D viewer had been hand-tuned with softer shadows, brighter exposure and a neutral environment light, but the script that rebuilds it before every release still produced the old flat version, so the next release would have quietly undone the tuning; the script now produces the tuned viewer, and the same settings apply to the fallback view for chat apps that cannot run the interactive one. The tool SDK's quick-start example pointed at an API that does not exist and skipped the wrapper an MCP server needs to return anything at all, so following it produced a tool that answered with nothing; it now runs end to end against a real free three.ws endpoint.</description></item>+ <item>+ <title>The Walk page shows the real top walkers again instead of claiming nobody has walked</title>+ <link>https://three.ws/walk</link>+ <guid isPermaLink="false">2026-08-15:The Walk page shows the real top walkers again instead of claiming nobody has walked</guid>+ <pubDate>Sat, 15 Aug 2026 12:00:00 GMT</pubDate>+ <category>fix</category>+ <description>The Top walkers board on /walk read "No walks logged yet. Be the first to put down some distance" for every visitor, while twenty-nine people had in fact logged distance and the full leaderboard at /walk-leaderboard listed all of them. The page was asking the leaderboard endpoint for the ranking and getting a correct answer back, then looking for it under half a dozen field names the endpoint has never used, finding nothing, and drawing its empty state. It now reads the shape the endpoint actually returns, so the top three walkers appear with their handles, avatars, distances, and a link to each profile, and it asks for only the three rows it draws instead of the first fifty. The social section under it now shows placeholder cards while it loads rather than a blank gap.</description>+ </item><item><title>Three docs pages stop pointing readers at a dead page, a missing section, and a snippet that does nothing</title><link>https://three.ws/docs/troubleshooting</link>@@ -22428,7 +22452,7 @@<guid isPermaLink="false">2026-06-05:Voice Lab</guid><pubDate>Fri, 05 Jun 2026 12:00:00 GMT</pubDate><category>launch</category>- <description>Browse about 850 voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.</description>+ <description>Browse hundreds of voices across Microsoft Edge, Gemini, NVIDIA, OpenAI and ElevenLabs, preview any of them, or clone your own from a short recording, then give it to your agent.</description></item><item><title>Walk: 3D Avatar AR</title>