Skip to content

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

fix(avatar-cli): restore the documented validate capture and publish the terminal accessibility repair

nirholas8ac258ce70d7
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 76b5f11b0..bfbd46bb4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
## 2026-08-16
+- **Docs · Avatar Artifact viewer** (`/docs/avatar-artifact`): The standalone avatar viewer you paste into any site as a one-line iframe: its ?agent= and ?model= parameters, the hosts a model may come from, how a clip-less rig still comes to life, and the sentence the page shows for every way it can fail.
- **A coin world link now unfurls as the live world, and only ever points at three.ws**: Pasting a coin community link into WordPress, Ghost, Discord, Notion or any editor built on embed.ly or iframely now turns it into the live 3D world instead of a plain link. The service that answers those editors existed but nothing on the page told them where to find it, so it was never called. Two problems in that service are fixed at the same time. It used to build the embed from the address in the incoming request rather than from the site address, so a crafted request could hand an editor an embed that framed somebody else site under the three.ws name; the embed is now always built from three.ws itself. And it used to answer for any link at all, returning a generic three.ws world card for a link that was never ours; it now says plainly that the link is not one of ours. It also answers in both of the formats the standard defines, and ignores a coin address that is not a real Solana address instead of putting it in the embed. Separately, the share images for pages and coin worlds, the holder leaderboard card, the free model verification check and the Oracle coin page now accept only read requests, and answer anything else with a clear message rather than doing the work anyway. (`/play`) `[fix, security]`
- **A dropped page in the agent directory no longer throws away everything you scrolled past**: The agent index loads in pages of 24, and until now a network hiccup on page five did more damage than the hiccup deserved. The Load more button vanished, and the only recovery on offer restarted the list from the top, discarding the hundred-odd agents already on screen. A failed page now says so on its own terms and offers Resume loading, which picks up exactly where it stopped; Start over is still there if you want it, and the error notice clears itself once the page arrives instead of lingering under the results. Three smaller repairs came with it: loading another page no longer redraws the agents already listed, so the grid stops flickering once a few pages are in; the search box and sort menu now show a proper focus ring for keyboard users instead of a one-pixel border tint; and the source filter chips reach a full-size tap target on tablets, not just on phones. (`/agents`) `[fix, improvement]`
- **A GitHub seed that stores nothing no longer costs you six hours**: Seeding an agent from your GitHub profile, repositories and READMEs is allowed once every six hours per agent. Two failures used to spend that window without storing a single fact. The first is entirely ours: when every model provider is busy at once, nothing is read and nothing is written, and you were still locked out for six hours over an outage you did not cause. The second is the run that comes back with no usable facts, which tells you to add a README or another repository, advice you could not act on until the window expired. Both now hand the window straight back, so you can try again as soon as a provider frees up or as soon as you have widened your selection, and the reply says which of the two happened. Nothing else changes: your existing memories are untouched by a failed run, the seed still reads only the profile, repositories and READMEs you ticked, and disconnecting GitHub still deletes every fact it wrote across all your agents. (`/docs/github-memory-seeding`) `[improvement, fix]`
@@ -18,6 +19,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
- **An expired X connection now offers to reconnect instead of sending you to the login page**: Posting a Walk screenshot or clip to X goes through your connected X account. When that connection could no longer be renewed, the share endpoint answered with the same code it uses for a signed-out visitor, so the capture panel told an already signed-in user to sign in to three.ws and linked them to the login page, which fixed nothing. That case now comes back as its own answer with the X connect link attached, and the panel offers a Reconnect X button that takes you straight to the flow that repairs it. The share caption also lost a stray dash. (`/walk`) `[fix]`
- **An unsubscribe now sticks, even if you click an old confirm link**: Every three.ws newsletter email carries a confirm link and an unsubscribe link that share one token, so a reader who unsubscribed and later opened an older email and clicked confirm was quietly put back on the list. That link no longer signs anyone up: it says you are unsubscribed and leaves it that way, and opting back in starts from the site. Three other fixes ship alongside it. The allocation planner at /meta-allocator no longer serves an empty basket when the model declines to answer, it falls back to the allocator that always produces a real spread across verified leaders. The motion capture uploader told you an upload link lasted ten minutes when it really lasted five, so a large video queued on that promise could fail with nothing explaining why. And an agent whose stablecoin payment is rejected by the x402 Bazaar now gets the Bazaar's own details back instead of a different server's, so its retry is aimed at the right place. (`/changelog`) `[fix]`
- **AR Studio: the language picker no longer sits on top of the dock controls**: On the AR Studio page the site-wide floating language picker landed in the bottom-right corner, directly over the right-hand end of the control dock, so the Clear button (and the buttons beside it on a phone) could not be tapped at all. The picker now lives in the studio's own top bar, where it is always reachable and leaves the dock its full width. It also shrinks instead of pushing its neighbours off screen on narrow phones, which fixes the same crowding anywhere else it appears in a tight header. The page description now also names every place models can come from: your forges, the community feed, the free CC0 object library, and any GLB link. (`/ar/studio`) `[fix, improvement]`
+- **Atlas: a long search term no longer knocks the page sideways, and the clear button is tappable**: /atlas is the whole platform as one searchable map, and its search box had three rough edges. Searching for a very long unbroken term (a pasted URL, a run-on word) pushed the text past the edge of its box and gave the entire page a horizontal scrollbar, so the layout slid sideways under you while you typed. Long text now wraps wherever it has to, and the same rule keeps long route paths from stretching the result grid on a narrow phone. The box also drew two clear buttons side by side, the one the browser paints for a search field and the one the page draws, which every other search surface on the site already suppresses and this one had missed. The remaining clear button was a 16 by 18 pixel target, well under the minimum anyone should have to hit on a phone. It is now a 26 pixel square with its own hover and keyboard focus treatment. (`/atlas`) `[fix, improvement]`
- **Autopilot Activity keeps up when you switch agents, and its receipts link to the exact memory**: Autopilot Activity lists every action your agents took on your behalf, and it has an agent picker for narrowing that list to one of them. Picking a second agent while the first was still loading used to do nothing at all: the picker and the address bar moved to the new agent, but the request for it was never sent and the page kept showing the previous agent's actions, with no sign anything had gone wrong. Whichever agent you picked last now wins, however fast you switch. Two more repairs: each receipt cites the memories that motivated the action, and those links now scroll to the cited memory and highlight it instead of dropping you at the top of a long Knowledge list; and when an agent has no actions yet, the page sends you to that agent's own autopilot controls rather than the dashboard, which is where the setting you need actually lives. (`/autopilot-activity`) `[fix, improvement]`
- **Avatar link previews stop rebuilding themselves, and two feeds answer a bad page link plainly**: An audit of the chat, media and board endpoints found three real breaks. When you share an avatar link, the preview image is drawn once and then kept, so every later share is instant. A mix-up between the stored image and its address meant every freshly drawn preview was mistaken for an old broken one: the saved copy was thrown away and the picture was redrawn from scratch on every single share. The preview is now kept, as it always should have been, and genuinely broken old records are repaired the next time they are drawn. Separately, the character directory answered a hand-edited or stale "next page" link with a server error instead of saying the value was not a timestamp, and the bounty board did the same when the page size in the address was not a number. Both now answer plainly, and the board simply falls back to its normal page size. (`/characters`) `[fix]`
- **Avatar Studio trait library: the mirror can no longer be pointed at someone else's server**: Avatar Studio loads its hair, clothing and accessory traits through a three.ws address that mirrors the open-source trait library, so your browser never talks to an outside CDN. The list of where those traits live was being written using the address in the incoming request, and that address can be set by whoever sends the request. Because the mirror is cached at the edge for a day, one crafted request could have left every Avatar Studio visitor loading trait files from a stranger's server for the next 24 hours. That list is now written the same way for everybody, with no part of it taken from the request, so there is nothing left to steer. Three smaller repairs went in alongside it: asking the mirror how big a file is no longer answers with the compressed size (a trait manifest claimed 2 KB and delivered 20 KB), the mirror will not hand back a web page or script even if the upstream library ever served one, and a failure at the library now answers plainly as either not found or mirror is down instead of passing its raw status through as ours. (`/avatar-studio`) `[security, fix]`
@@ -35,6 +37,7 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
- **Live stage hosts stop swallowing hyphens, and only confirmed tippers get greeted by name**: Two repairs to the Living Stages host. The cleanup pass that tidies the host's spoken line was stripping hyphens along with the invisible characters it was meant to remove, so every show heard "on chain" where the host said "on-chain" and "state of the art" for "state-of-the-art"; punctuation now survives and only invisible characters are removed. Separately, the host's memory of returning regulars counted tips whose on-chain settlement had not been confirmed yet, which meant anyone could post an unconfirmed tip under a name of their choosing and hear the host greet it aloud. That memory now reads confirmed tips only, matching the public leaderboard. Smaller repairs in the same pass: a stage that has not opened its first show is no longer cached as empty at the edge, an over-long coin address is refused instead of quietly trimmed into a different plaza, and the running tip total is written before the response returns. (`/stage`) `[fix, security]`
- **Memory seeding reads Farcaster with no vendor key, and X now works from the app credentials we already hold**: The memory-seeding demo asks a GitHub, X or Farcaster handle what it can learn about you in public, then writes a short memory seed your agent can carry. Two of those three connectors were showing a "connector not configured" card to everyone, because each insisted on a vendor key nobody had set. Farcaster now reads through the same failover the rest of the platform uses: an indexer when a key exists, and a public Farcaster hub with no key at all otherwise, so it returns real profiles and casts on any deployment. Anything the keyless lane genuinely cannot know, such as follower counts, is now shown as unknown instead of zero. The X connector accepts the app key and secret the platform already stores, exchanging them for a read-only token itself, so it no longer needs a separate token provisioned by hand. A mistyped Farcaster name now comes back as not found rather than as an upstream outage, and all four endpoints are documented in the API reference for the first time. (`/demos/memory-seed`) `[fix, improvement, docs]`
- **Mistyped filters on the revenue, checkout and ring endpoints answer plainly instead of failing**: The public revenue feed, the checkout SKU API and the ring report all took values straight from the URL into the database. A timestamp that was not a timestamp, a checkout id that was not an id, and the query fell over: the caller got an opaque "something went wrong" with a support reference, and our own alerting logged a fault that was only ever a typo. All three now say what was wrong, in words, with the right status code. The ring report also stops treating an unrecognised period as "show me everything", which quietly answered the wrong question and ran the most expensive query it has. The revenue feed now echoes back the endpoint filter it actually applied rather than the raw text it was sent, and the paid-call lookup only accepts something shaped like a real transaction signature. (`/x402-revenue`) `[fix]`
+- **Multi-LLM Brain now shows every model it can actually run, and says so when one is out**: The /brain playground kept its own list of models on the page, and it had drifted from the models the server can really reach: nine were missing from the picker, including the current Claude Opus 5 and Sonnet 5 and all seven free models a signed-out visitor is allowed to use, while two entries in the list no longer existed at all. The picker is now built from the live model service, so it always matches what can be sent. A signed-out visitor opens on four models that are genuinely free to use instead of paid ones that could only refuse, and every model that needs an account is marked with a lock and a sign-in link rather than failing after you press Send. When the model list cannot be loaded the page now says so and offers a Retry that recovers, instead of quietly offering every model and erroring on send. When a model fails after every fallback route, the column now explains what happened and what to try next instead of printing a bare upstream word like Forbidden. Model choices are remembered between visits. On a phone the session list is reachable again through a drawer, the compare columns snap one per screen, and the Send button is no longer covered by the floating helper widgets. Model chips are proper buttons with keyboard focus, every field has a label, and clearing your persona from the sidebar now clears it everywhere. (`/brain`) `[improvement, fix]`
- **Paying with $THREE works again: credits top-ups, tips and purchases stopped failing before they were signed**: Every attempt to top up credits with $THREE died in your wallet with a message about an incorrect program id, before you could approve anything and before any money moved. $THREE is issued under Solana's newer token standard, Token-2022, but the pages building the transfer were addressing the older one. That points the transfer at an account that does not exist, so the network rejected it during the safety check your wallet runs first. The credits page, agent tipping and the in-page purchase flow now read the standard directly from the coin itself and build against whichever one it actually uses, so $THREE and ordinary SPL tokens both go through. Two related touches on the credits page: it now tells you up front if the connected wallet holds no $THREE, or not enough for the amount you typed, instead of letting the wallet fail with something cryptic. The same mix-up was quietly hobbling the treasury side: the buyback and micro-buy lanes read their own $THREE balance from the wrong place, always saw zero, and so never moved what they had bought into the treasury. They now read and move it correctly. (`/credits`) `[fix]`
- **Prompt polish keeps its quotes out of your model, and a busy Forge says so plainly**: Three fixes across the Forge endpoints. The Enhance button, which rewrites your rough idea into the prompt the generator actually draws from, stopped removing the quotation marks the writing model wraps around its answer: those quotes were being handed straight to the image step as if you had asked for them to be drawn. Both plain and curly quotes are stripped again. Second, when you ask the studio to change a model you already made and the generator is saturated, you now get an honest "busy, try again in a moment" with a wait time instead of a vague failure that blamed the wording of your request. Third, the self-hosted generator panel now tells a caller which input was missing rather than reporting the whole feature as unconfigured, and its status light no longer claims nothing is set up when the only problem was the address you typed. (`/forge`) `[fix, improvement]`
- **Robinhood Chain worlds survive a hiccup in the live feed instead of going dark**: The in-world trading terminal inside a Robinhood Chain world polls the chain feed every five seconds, and the Worlds lobby lists those coins from the same feed. A single event that arrived without its payload, or with a timestamp the chain wrote in an unexpected format, took both requests down with a server error: the terminal showed a dead tape, and the lobby's Robinhood Chain tab quietly emptied with nothing on screen to say why. Bad events are now skipped and the rest of the tape still renders, a price that is genuinely missing stays blank rather than printing as a real $0 trade, and a launch with no readable contract address no longer produces a card that leads nowhere. Asking for trades with an address that is not a contract address now gets a clear error back instead of an empty list that read as "this coin has never traded". (`/worlds`) `[fix]`
@@ -56,10 +59,15 @@ Public history for [three.ws](https://three.ws), newest first. New pages come fr
- **The AGI page no longer goes quiet when its 3D body cannot load**: The AGI page is built around a live 3D body that reacts to the market, and when that model could not be fetched the page said nothing at all. The loading line disappeared on a timer whether or not anything had arrived, so the centrepiece became a silent empty panel with no explanation and no way to try again, while the decision stream and track record beside it carried on as normal. A body that fails to load now falls back to the platform's own avatar, so there is still something on the stage reacting to the agent's mood. If that cannot be fetched either, the stage says so plainly, makes clear that the agent is still trading and that the decisions and track record below are live and unaffected, and offers a button to try embodying again. The loading line now stays up until the body genuinely arrives instead of vanishing early, and the stage no longer re-requests a model it already knows is unreachable every twenty seconds. The page's description shown in search results and link previews was also carrying an older wording than the page itself. (`/agi`) `[fix, improvement]`
- **The Animation Gallery keeps its keyboard, its focus, and its promises about what will play**: The gallery of nearly 3,000 motion clips had a set of small breakages that added up. Roughly a third of the mocap catalogue has no picture made for it yet, and the page was asking the picture store for one anyway: every one of those requests came back empty, so a third of the cards spent a request to arrive at the same drawing they now show immediately. Opening a clip's details trapped nobody: pressing Tab walked out of the open window and into the page behind it, and closing the window dropped you back at the very top of the page instead of the card you came from. The playback controls were on show while a preview was still loading and stayed on show when it failed, so there were buttons that did nothing. Arrow keys jumped to the next clip even while you were dragging the speed slider, which moved the slider and changed the clip at the same time. Sharing a filtered view was misleading too: a link carrying the loops or one-shot filter applied it but left the buttons showing All, and a link with a filter the page does not recognise emptied the sort menu. All of it is fixed, along with the keyboard hint the window shows: Space really does play and pause now, from the moment the window opens. (`/animations`) `[fix, improvement]`
- **The arbitrage board says which price feeds it actually reached**: The x402 Arbitrage page compares what different providers charge for the same capability, and it builds that comparison from several facilitator catalogs at once. When one of those catalogs was unreachable, its listings simply vanished from every comparison and the page said nothing about it. A scan that had only seen half the market looked exactly like a market with fewer price gaps in it, which is the opposite of what the page is for. Each scan now names the facilitators it reached and how many listings each one contributed, and calls out any that dropped out along with the reason, so a short board is never mistaken for a quiet one. Separately, closing the payment window without paying used to leave the Pay cheapest button dead for four seconds before it would accept another click. Deciding not to pay now hands the button straight back. (`/arbitrage`) `[fix, improvement]`
+- **The Arena now ranks agents that actually traded above the ones that sat out**: A tournament scored on realized P&L was handing a clean zero to every entrant that never opened a trade, which put them above every agent that took a real loss. On today's Daily Arena that meant the eight agents who did nothing held the whole podium while the two that traded sat at the bottom, and because a no-trade entrant can never clear the prize gates, the splits would have paid out nothing at all. Ranking now puts a result ahead of no result: entrants with no closed trade in the window sit below everyone who competed. They stay on the board with their honest zero, they just no longer outrank the field. The trophy, the podium on the front page and the prize ladder now only ever name an entrant that has actually traded, and the ladder only names an agent that is genuinely prize-eligible instead of promising a payout the settlement would never send. A tournament nobody has traded in now says it is awaiting the first ranked trade rather than crowning a zero. Also fixed: the tournament name in the browser tab was being reset to the generic page title a second after the page loaded, and the create-tournament form's fields overflowed their sheet on a 320px phone. (`/arena`) `[fix, improvement]`
- **The artifact builder now reports what it actually measures, and stops flagging legal artifacts as over budget**: The /artifact page is the pre-flight for pasting a three.ws agent into a Claude conversation: it builds the self-contained HTML, renders it in a sandbox that mirrors Claude's, and reports what you are about to paste. Two of those numbers were misleading. The stat labelled First paint was really the time the endpoint took to hand over the bytes; nothing had been rendered yet when it was measured. It is now split into Fetch, which is that download, and Render, which times how long the sandboxed frame takes to parse the document and start the viewer. The size gauge was grading the finished artifact against the size limit for the model inside it, but inlining base64-encodes the model and costs four bytes for every three, so a perfectly acceptable 7 MB artifact was painted amber and anything built from a model over about 5.6 MB went red, contradicting the endpoint that had just accepted it. Size is now graded against the roughly 8.6 MB an accepted model can actually produce, and hovering the number tells you how much of that budget you have used. The Size budget note explained the same thing incorrectly and now spells out the base64 cost. (`/artifact`) `[fix, improvement]`
-- **The Avatar CLI page now shows real command output, and its terminal is readable by a screen reader**: The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and one of its examples was staged rather than captured: the validate card showed a seven-error failure stitched together from errors no single manifest produces. It now shows a verbatim run against a genuinely broken manifest, with the instance paths (/, /mesh, /skeleton) the CLI really prints, so what you read on the page is what your shell will print back. The replayed terminal in the hero was marked up as one flat image, which meant a screen reader announced a one-line caption and hid both the transcript and the Replay button inside it; the transcript is now a named, readable region and Replay is an ordinary button again. The live viewer's failure notice keeps a link to the web component reference, and that link now looks like a link and shows a focus ring when you tab to it, as do the scrollable code blocks. The copy button's fallback hint used to tell everyone to press Command C; it now names the shortcut for the keyboard you are actually on. (`/avatar-cli`) `[fix, improvement]`
+- **The ASL Alphabet drill now answers every press, and its package link goes somewhere real**: Two things on /asl-alphabet did nothing when you used them. The Check button in the reading drill sat enabled from the moment the page loaded, but pressing it before starting a round, with an empty box, or after revealing an answer changed nothing on screen at all, which reads as a broken button. Check now always replies: it tells you to press start when there is no round to grade, tells you to type what you read when the box is empty, and says the round is finished once it has been graded or revealed. The start button's label follows the drill instead of being stuck on "Skip to the next one" after a round ends, so the feedback line can name the button you actually need. A wrong answer no longer says "Watch it again" when the round is already closed; it points at "Show it again" to study the shape and then move on. The drill's feedback is now announced to screen readers instead of changing silently. Separately, the note at the foot of the page (and the matching one on /sign-mirror) linked the signing engine to an npm page that does not exist: the package is in the repository and its registry publish has not landed yet, so both now link to the real source and say how to install it. (`/asl-alphabet`) `[fix, improvement]`
+- **The assistant widget now tells you the truth when its chat lane is busy**: Two things on the assistant widget page reported the wrong thing. The live preview status line always read "Starting the live preview...", even once the avatar was up and running, because the translation pass reset it a moment after the page wrote it. Every state now carries its own translated line, so the preview says what it is actually doing, and it reports a real failure with a Retry button when the widget or its avatar does not come up instead of silently claiming success. Second, when the free chat lane was throttled the widget only said "Chat failed (HTTP 503)", and a signed-out visitor could even be told to sign in for a capacity problem the widget has no sign-in for. The chat endpoint now answers with a proper at-capacity result and how long to wait, and the widget shows that in full and points at the settings panel where a visitor can drop in their own Groq or OpenRouter key. Editing the builder no longer re-downloads the avatar on every keystroke. (`/assistant`) `[fix, improvement]`
+- **The Avatar Artifact page now shows real avatars, not just its house portrait**: /avatar-artifact is the standalone viewer you can paste into any site as a bare iframe, but it only ever rendered one built-in procedural figure, so there was no way to point it at an actual avatar. It now takes ?agent=<id> to show any public agent's avatar, resolved live from the platform, or ?model=<glb url> to show a model directly from an allowed asset host. The agent's name becomes the heading and an Open profile link appears, the Copy embed button hands you an iframe of exactly what you are looking at, and avatars with no baked animation play the shared idle clip retargeted onto their own skeleton instead of standing in a T-pose. Every way it can fail now says what happened and what to do next: an id that matches no agent, an agent whose avatar is private or not created yet, a model URL from a host the viewer will not render, a model that would not load, and a browser without WebGL. Loading a large model reports its download progress instead of guessing, and a slow but healthy download is no longer cut off as a failure. With no parameters the page keeps its original portrait. (`/avatar-artifact`) `[feature, improvement, fix]`
+- **The Avatar CLI page terminal is readable by a screen reader, and its recovery link is reachable by keyboard**: The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and the replayed terminal in its hero was marked up as one flat image. A screen reader announced a single caption for it and hid everything inside: the transcript itself, and the Replay button that restarts it. The transcript is now a named region a screen reader can read line by line, Replay is an ordinary button again, and the transcript reports itself as busy while it is still typing rather than being read half-written. Below it, the live viewer proves the embed snippet works, and when the component loader cannot be reached it swaps in a notice with a link to the web component reference. That link is the only way out of the failure, and it neither looked like a link nor showed a focus ring when tabbed to; it now does both. The scrollable code blocks, which the browser makes keyboard focusable, now show the site focus ring instead of a hairline default. And the copy button's clipboard fallback used to tell everyone to press Command C, including on Windows and Linux; it now names the shortcut for the keyboard you are actually on. (`/avatar-cli`) `[fix, improvement]`
- **The AWS page stops promising AWS invoicing three.ws does not do, and the GLB optimizer reads compressed avatars again**: Three corrections found by reading the docs pages against the code they describe. The AWS page told you a three.ws subscription would bill usage onto your AWS invoice, draw down AWS credits, and count toward Enterprise Discount Program commitments. None of that is the plan: the Marketplace subscription is free, and usage is paid per call in USDC over x402, exactly as it is for any other caller, so there is nothing to reconcile on your AWS bill at all. The page now says that, and describes the free subscription, the single key that opens every x402 endpoint, and the absence of any trial clock. Separately, the GLB optimization command documented on the 3D asset pipeline page failed on every compressed avatar, which is most of them, because it loaded the glTF extensions without the matching decoders; it now decodes meshopt and Draco and leaves the output as plain glTF the way its own contract promises. The NVIDIA models page also listed the tool-capable fallback model under the wrong vendor prefix. (`/aws`) `[fix, docs]`
- **The background agent mint says out loud whether it ran**: Every avatar that finishes a scan or a prompt is given a draft on-chain identity in the background. That step is deliberately best-effort, so it can never cost you your avatar, but it was also completely silent: whether it minted, was skipped because the deployment holds no minting key, or was switched off entirely, the logs looked identical, which is how a run of avatars can go by without anyone noticing the identity was never created. Each finished avatar now records the outcome of both chains, including the reason a leg was skipped. The documentation for the feature also gained a way to prove the whole path on a machine that cannot reach the public test-network faucet, which is routinely exhausted on shared build machines: run your own node with the real Metaplex programs, and the proof reports which node served it rather than handing back an explorer link that leads nowhere. (`/docs/draft-agent-mint`) `[improvement, docs]`
+- **The BNB block race stops showing a dead chain's last number as if it were live**: The live block race compares BNB Chain, Base, Ethereum and Solana by measuring each one off a public RPC every few seconds, and it promises that a chain going quiet says so instead of freezing a fake number. The four lane cards kept that promise; the big headline number did not. When the BNB RPC stopped answering, the headline held its last reading at full brightness, kept claiming it was N times faster than Base, and the Updated stamp sat on "just now" no matter how long the outage ran. The headline now dims the moment the reading stops being live, labels it as last measured and prints its real age in seconds, and drops the speed comparison rather than race a stale number against a live one. The Updated stamp ages off the measurement time the server reports, and a poll that measured nothing no longer refreshes it. If every chain goes down at once the page now raises its outage notice and its Retry button, which previously could only ever appear before the first successful reading and so was unreachable during a real outage. Retry now shows it is working, makes a genuinely new request rather than reporting an older in-flight one's failure, and no longer sits underneath the floating helper widgets where the click was being swallowed. A lane that has never returned a sample says so plainly instead of pointing at a reading that is not on screen, each poll goes to the network instead of replaying a cached body up to fifteen seconds old, and screen readers now hear a lane announce a change of state rather than re-reading all four numbers twelve times a minute. (`/bnb-latency`) `[fix, improvement]`
- **The BNB Chain page now says how many demos are actually live**: The BNB Chain page checks each of its three demo tracks against the running deployment and reports the tally in its header. That line was stuck: it read "Checking which tracks are live" forever, even once all three cards beside it had finished and said Live. The page's own result was being written first and then overwritten by the translation pass that swaps the page into your language, so the finished count never survived. The header now holds its result and reads "3 of 3 demo tracks live now". The same checks were also leaving four cancelled requests in the browser's network log on every visit, which made a healthy page look like it was failing to anyone with developer tools open. They now close cleanly. (`/bnb`) `[fix]`
- **The creations gallery now counts every creation a creator has made, and pages all the way to the end**: The gallery ranked creators and paged the feed over only the newest 120 creations, however many the feed actually held. A creator's older work was invisible in their profile count, the top-creator board scored people on a fraction of their output, and scrolling the gallery dead-ended early with items still left to show. It now reads the full feed depth it always meant to, and does it in two database round trips instead of two per item, so the deeper read loads faster than the shallow one did. (`/creations`) `[fix, improvement]`
- **The FLAME paper link in the Avatar Engines Atlas opens the paper again**: FLAME is the parametric head model underneath most of the photoreal-head engines in the atlas at /avatar-engines, so its Paper button is one a builder actually follows. It pointed at a Max Planck publication page that answers every visitor, browsers included, with a 403 Forbidden, so the button led to an error page instead of the paper. It now opens the paper PDF that the FLAME project site itself links to. All 68 outbound links on the page were re-checked afterwards and every one of them resolves. (`/avatar-engines`) `[fix]`
diff --git a/data/changelog.json b/data/changelog.json
index 5066aefe6..2c87dd9ce 100644
--- a/data/changelog.json
+++ b/data/changelog.json
@@ -1,6 +1,77 @@
{
"$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": "2026-08-16",
+ "title": "Atlas: a long search term no longer knocks the page sideways, and the clear button is tappable",
+ "summary": "/atlas is the whole platform as one searchable map, and its search box had three rough edges. Searching for a very long unbroken term (a pasted URL, a run-on word) pushed the text past the edge of its box and gave the entire page a horizontal scrollbar, so the layout slid sideways under you while you typed. Long text now wraps wherever it has to, and the same rule keeps long route paths from stretching the result grid on a narrow phone. The box also drew two clear buttons side by side, the one the browser paints for a search field and the one the page draws, which every other search surface on the site already suppresses and this one had missed. The remaining clear button was a 16 by 18 pixel target, well under the minimum anyone should have to hit on a phone. It is now a 26 pixel square with its own hover and keyboard focus treatment.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/atlas"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "The ASL Alphabet drill now answers every press, and its package link goes somewhere real",
+ "summary": "Two things on /asl-alphabet did nothing when you used them. The Check button in the reading drill sat enabled from the moment the page loaded, but pressing it before starting a round, with an empty box, or after revealing an answer changed nothing on screen at all, which reads as a broken button. Check now always replies: it tells you to press start when there is no round to grade, tells you to type what you read when the box is empty, and says the round is finished once it has been graded or revealed. The start button's label follows the drill instead of being stuck on \"Skip to the next one\" after a round ends, so the feedback line can name the button you actually need. A wrong answer no longer says \"Watch it again\" when the round is already closed; it points at \"Show it again\" to study the shape and then move on. The drill's feedback is now announced to screen readers instead of changing silently. Separately, the note at the foot of the page (and the matching one on /sign-mirror) linked the signing engine to an npm page that does not exist: the package is in the repository and its registry publish has not landed yet, so both now link to the real source and say how to install it.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/asl-alphabet"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "The BNB block race stops showing a dead chain's last number as if it were live",
+ "summary": "The live block race compares BNB Chain, Base, Ethereum and Solana by measuring each one off a public RPC every few seconds, and it promises that a chain going quiet says so instead of freezing a fake number. The four lane cards kept that promise; the big headline number did not. When the BNB RPC stopped answering, the headline held its last reading at full brightness, kept claiming it was N times faster than Base, and the Updated stamp sat on \"just now\" no matter how long the outage ran. The headline now dims the moment the reading stops being live, labels it as last measured and prints its real age in seconds, and drops the speed comparison rather than race a stale number against a live one. The Updated stamp ages off the measurement time the server reports, and a poll that measured nothing no longer refreshes it. If every chain goes down at once the page now raises its outage notice and its Retry button, which previously could only ever appear before the first successful reading and so was unreachable during a real outage. Retry now shows it is working, makes a genuinely new request rather than reporting an older in-flight one's failure, and no longer sits underneath the floating helper widgets where the click was being swallowed. A lane that has never returned a sample says so plainly instead of pointing at a reading that is not on screen, each poll goes to the network instead of replaying a cached body up to fifteen seconds old, and screen readers now hear a lane announce a change of state rather than re-reading all four numbers twelve times a minute.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/bnb-latency"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "The Arena now ranks agents that actually traded above the ones that sat out",
+ "summary": "A tournament scored on realized P&L was handing a clean zero to every entrant that never opened a trade, which put them above every agent that took a real loss. On today's Daily Arena that meant the eight agents who did nothing held the whole podium while the two that traded sat at the bottom, and because a no-trade entrant can never clear the prize gates, the splits would have paid out nothing at all. Ranking now puts a result ahead of no result: entrants with no closed trade in the window sit below everyone who competed. They stay on the board with their honest zero, they just no longer outrank the field. The trophy, the podium on the front page and the prize ladder now only ever name an entrant that has actually traded, and the ladder only names an agent that is genuinely prize-eligible instead of promising a payout the settlement would never send. A tournament nobody has traded in now says it is awaiting the first ranked trade rather than crowning a zero. Also fixed: the tournament name in the browser tab was being reset to the generic page title a second after the page loaded, and the create-tournament form's fields overflowed their sheet on a 320px phone.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/arena"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "Multi-LLM Brain now shows every model it can actually run, and says so when one is out",
+ "summary": "The /brain playground kept its own list of models on the page, and it had drifted from the models the server can really reach: nine were missing from the picker, including the current Claude Opus 5 and Sonnet 5 and all seven free models a signed-out visitor is allowed to use, while two entries in the list no longer existed at all. The picker is now built from the live model service, so it always matches what can be sent. A signed-out visitor opens on four models that are genuinely free to use instead of paid ones that could only refuse, and every model that needs an account is marked with a lock and a sign-in link rather than failing after you press Send. When the model list cannot be loaded the page now says so and offers a Retry that recovers, instead of quietly offering every model and erroring on send. When a model fails after every fallback route, the column now explains what happened and what to try next instead of printing a bare upstream word like Forbidden. Model choices are remembered between visits. On a phone the session list is reachable again through a drawer, the compare columns snap one per screen, and the Send button is no longer covered by the floating helper widgets. Model chips are proper buttons with keyboard focus, every field has a label, and clearing your persona from the sidebar now clears it everywhere.",
+ "tags": [
+ "improvement",
+ "fix"
+ ],
+ "link": "/brain"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "The assistant widget now tells you the truth when its chat lane is busy",
+ "summary": "Two things on the assistant widget page reported the wrong thing. The live preview status line always read \"Starting the live preview...\", even once the avatar was up and running, because the translation pass reset it a moment after the page wrote it. Every state now carries its own translated line, so the preview says what it is actually doing, and it reports a real failure with a Retry button when the widget or its avatar does not come up instead of silently claiming success. Second, when the free chat lane was throttled the widget only said \"Chat failed (HTTP 503)\", and a signed-out visitor could even be told to sign in for a capacity problem the widget has no sign-in for. The chat endpoint now answers with a proper at-capacity result and how long to wait, and the widget shows that in full and points at the settings panel where a visitor can drop in their own Groq or OpenRouter key. Editing the builder no longer re-downloads the avatar on every keystroke.",
+ "tags": [
+ "fix",
+ "improvement"
+ ],
+ "link": "/assistant"
+ },
+ {
+ "date": "2026-08-16",
+ "title": "The Avatar Artifact page now shows real avatars, not just its house portrait",
+ "summary": "/avatar-artifact is the standalone viewer you can paste into any site as a bare iframe, but it only ever rendered one built-in procedural figure, so there was no way to point it at an actual avatar. It now takes ?agent=<id> to show any public agent's avatar, resolved live from the platform, or ?model=<glb url> to show a model directly from an allowed asset host. The agent's name becomes the heading and an Open profile link appears, the Copy embed button hands you an iframe of exactly what you are looking at, and avatars with no baked animation play the shared idle clip retargeted onto their own skeleton instead of standing in a T-pose. Every way it can fail now says what happened and what to do next: an id that matches no agent, an agent whose avatar is private or not created yet, a model URL from a host the viewer will not render, a model that would not load, and a browser without WebGL. Loading a large model reports its download progress instead of guessing, and a slow but healthy download is no longer cut off as a failure. With no parameters the page keeps its original portrait.",
+ "tags": [
+ "feature",
+ "improvement",
+ "fix"
+ ],
+ "link": "/avatar-artifact"
+ },
{
"date": "2026-08-16",
"title": "AR Studio: the language picker no longer sits on top of the dock controls",
@@ -23,8 +94,8 @@
},
{
"date": "2026-08-16",
- "title": "The Avatar CLI page now shows real command output, and its terminal is readable by a screen reader",
- "summary": "The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and one of its examples was staged rather than captured: the validate card showed a seven-error failure stitched together from errors no single manifest produces. It now shows a verbatim run against a genuinely broken manifest, with the instance paths (/, /mesh, /skeleton) the CLI really prints, so what you read on the page is what your shell will print back. The replayed terminal in the hero was marked up as one flat image, which meant a screen reader announced a one-line caption and hid both the transcript and the Replay button inside it; the transcript is now a named, readable region and Replay is an ordinary button again. The live viewer's failure notice keeps a link to the web component reference, and that link now looks like a link and shows a focus ring when you tab to it, as do the scrollable code blocks. The copy button's fallback hint used to tell everyone to press Command C; it now names the shortcut for the keyboard you are actually on.",
+ "title": "The Avatar CLI page terminal is readable by a screen reader, and its recovery link is reachable by keyboard",
+ "summary": "The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and the replayed terminal in its hero was marked up as one flat image. A screen reader announced a single caption for it and hid everything inside: the transcript itself, and the Replay button that restarts it. The transcript is now a named region a screen reader can read line by line, Replay is an ordinary button again, and the transcript reports itself as busy while it is still typing rather than being read half-written. Below it, the live viewer proves the embed snippet works, and when the component loader cannot be reached it swaps in a notice with a link to the web component reference. That link is the only way out of the failure, and it neither looked like a link nor showed a focus ring when tabbed to; it now does both. The scrollable code blocks, which the browser makes keyboard focusable, now show the site focus ring instead of a hairline default. And the copy button's clipboard fallback used to tell everyone to press Command C, including on Windows and Linux; it now names the shortcut for the keyboard you are actually on.",
"tags": [
"fix",
"improvement"
diff --git a/pages/avatar-cli.html b/pages/avatar-cli.html
index 10d622ece..ade29d8b2 100644
--- a/pages/avatar-cli.html
+++ b/pages/avatar-cli.html
@@ -189,10 +189,11 @@
gate rather than a report.
</p>
<pre class="cli-snip"><code>$ three-ws-avatar validate broken.json
-<span class="e">✖</span> broken.json is invalid <span class="dim">(3 errors)</span>
- <span class="e">•</span> <span class="c">/</span> must have required property 'createdAt'
- <span class="e">•</span> <span class="c">/mesh</span> must have required property 'sha256'
- <span class="e">•</span> <span class="c">/skeleton</span> must be equal to one of the allowed values</code></pre>
+<span class="e">✖ broken.json is invalid (7 errors)</span>
+ • / must have required property 'mesh'
+ • / must have required property 'skeleton'
+<span class="dim"> …</span>
+ • /id must match a schema in anyOf</code></pre>
</article>
<article class="cli-cmd-card">
diff --git a/prompts/swarm-100/route-avatar-cli.md b/prompts/swarm-100/route-avatar-cli.md
deleted file mode 100644
index 7b7e43a7b..000000000
--- a/prompts/swarm-100/route-avatar-cli.md
+++ /dev/null
@@ -1,119 +0,0 @@
-# Route audit: /avatar-cli (Avatar CLI)
-
-How to run: paste this file's repo path into a fresh Claude Code chat in this
-repository and say "run this work order". This file is fully self-contained:
-it depends on no other prompt file anywhere. If neighboring files in
-prompts/swarm-100/ are gone, that work is done; if present, ignore them.
-Every claim below rots; step 0 re-measures, and what you measure wins.
-
-## Operating clause (binding)
-
-- Read CLAUDE.md first. Its rules override everything, including this file.
-- Finish 100% in this session. Never end the turn with a question, an option
- list, or an unexecuted plan. A judgment call goes in one line of the final
- report; it never becomes a question that halts work.
-- The only permitted stops are the CLAUDE.md stop-and-ask gates: spending real
- funds or any irreversible on-chain write, git push or a production deploy,
- committing content that references a crypto project other than $THREE, and
- destroying unrecoverable data.
-- No mocks, no fake data, no placeholder stubs, no unfinished-work markers, no
- commented-out code. Real APIs and real integrations only.
-- The em-dash and en-dash characters are banned in everything you write.
-- Concurrent agents share this worktree: stage explicit paths only (never a
- bare add-everything), and commit finished work promptly.
-- Before committing, run: npm run check:rules -- --paths <files you touched>.
- It must exit 0.
-
-## Step 0: re-derive the current state
-
-1. This route's declared entry (path, description, added date):
-
- node -e "const p=require('./data/pages.json');for(const s of p.sections)for(const g of s.pages)if(g.path==='/avatar-cli')console.log(JSON.stringify(g,null,2))"
-
-2. Find the serving implementation: grep "/avatar-cli" in vercel.json (the live
- route table server/index.mjs boots from), then list matching files in
- pages/ and src/. STRUCTURE.md maps every surface to its directory if the
- grep is ambiguous.
-3. Production right now:
-
- curl -so /dev/null -w '%{http_code}\n' https://three.ws/avatar-cli
-
-4. Start (or reuse) the local dev server: npm run dev (vite, port 3000).
-
-## Task: audit this one route end to end, then fix everything found
-
-Route: /avatar-cli ("Avatar CLI", section "build").
-Declared purpose: The @three-ws/avatar-cli package: scaffold a hash-anchored avatar manifest from a wallet and a GLB, validate it against the schema, and print an embed snippet. Offline, deterministic, and safe to run in a release script.
-
-Drive the page in a real headless browser. Playwright is already a dev
-dependency; keep your probe script in the session scratchpad, never in the
-repo.
-
-1. Load http://localhost:3000/avatar-cli. Record the HTTP status, every console
- error and warning, and every failed network request.
-2. Exercise it like a first-time user: click the primary actions, run the
- primary flow if one exists, follow the page's internal links. Every button
- must do something real; every link must resolve. If the page renders data,
- prove in the network log that it came from a live endpoint, not a baked-in
- fallback array.
-3. States: verify populated, empty, loading, and error states are designed and
- reachable. Block the page's primary fetch in Playwright and confirm an
- actionable error state appears, not a blank void or a raw exception.
-4. Responsive: render at 320, 768, and 1440 px. No horizontal page scroll, no
- overlapping or clipped controls.
-5. Accessibility basics: one h1; interactive elements are real buttons or
- links with visible focus states; images carry alt text; form fields have
- labels.
-6. Head: a specific title tag and meta description that match the declared
- purpose above.
-7. Fix every defect at its root in the page's real source files and re-run the
- probe until clean. Defects in shared modules are in scope when this route
- exercises them.
-
-## Definition of done
-
-- [ ] Local load of /avatar-cli: HTTP 200, zero console errors, zero unhandled
- failed requests (a third-party outage must land in a designed fallback,
- not a broken page).
-- [ ] No dead interactive element on the page: no stub href="#", no button
- without a handler, no link to a missing path.
-- [ ] Empty, loading, error, and populated states verified.
-- [ ] 320 / 768 / 1440 px renders verified clean.
-- [ ] Title and meta description present and specific.
-- [ ] npm test passes; npm run check:rules -- --paths <touched files> exits 0.
-- [ ] Production status from step 0 recorded in the report; if production
- needs a deploy to pick up a fix, the report says so (deploys stay
- owner-gated).
-
-## Never blocked
-
-| Blocker | Resolution (act, do not ask) |
-|---|---|
-| Missing env var or credential | Check .env and .env.local, then the Cloud Run service env, then Secret Manager (the CLAUDE.md self-unblock playbook has the exact commands). If it truly exists nowhere, wire the code fully behind the env var, prove the wiring with a dry run, and name the single missing var in the report. |
-| gcloud not on PATH | Run: export PATH="$HOME/google-cloud-sdk/bin:$PATH" |
-| Dev server port 3000 busy | Another agent may be serving this repo; probe it and reuse if so. Otherwise start your own on a free port: npx vite --port 3101 |
-| Playwright browser missing | npx playwright install chromium |
-| A surface needs a signed-in user | Register a fresh account through the real /register flow against the real API and use it. Never mock the session. |
-| A defect sits in code you did not touch | Fix it if it blocks a Definition of done line (root cause it, never mask it). Otherwise note it in the report and continue. |
-| An unrelated test is red | Same rule. Never pipe npm test through tail; it masks exit codes. |
-
-## Close out (required)
-
-1. Verify every Definition of done line with the actual command output in
- front of you. Never claim a line you did not verify.
-2. Commit with explicit paths and a subject that describes the diff (house
- style: type(scope): what changed and why a reader cares).
-3. Delete this prompt file in that same commit:
-
- git rm prompts/swarm-100/route-avatar-cli.md
-
- The shrinking directory is this campaign's only progress ledger; there is
- no progress log to update.
-4. Final report: what step 0 measured, what changed (file list), evidence per
- Definition of done line, and any one-line judgment calls. No trailing
- questions, no unexecuted plans.
-
-If a line genuinely cannot pass inside this session (an external party must
-respond, or an owner-gated action is the final step), finish everything else,
-leave this file in place, and state exactly which line remains and who owns
-it. Never delete this file on a partial.
diff --git a/public/avatar-cli.css b/public/avatar-cli.css
index 652f12040..1544c6a38 100644
--- a/public/avatar-cli.css
+++ b/public/avatar-cli.css
@@ -490,7 +490,6 @@
.cli-snip .s { color: var(--success); }
.cli-snip .n { color: var(--warn); }
.cli-snip .e { color: var(--danger); }
-.cli-snip .c { color: var(--accent); }
.cli-snip .ok { color: var(--success); }
.cli-snip .dim { color: var(--ink-faint); }
diff --git a/public/changelog.json b/public/changelog.json
index 46dd0022b..a934d0062 100644
--- a/public/changelog.json
+++ b/public/changelog.json
@@ -1,11 +1,21 @@
{
- "generated_at": "2026-08-16T20:37:12.371Z",
+ "generated_at": "2026-08-16T21:09:53.836Z",
"generated_by": "scripts/build-page-index.mjs from data/pages.json + data/changelog.json",
"site": {
"name": "three.ws",
"url": "https://three.ws"
},
"entries": [
+ {
+ "date": "2026-08-16",
+ "type": "launch",
+ "title": "Docs · Avatar Artifact viewer",
+ "summary": "The standalone avatar viewer you paste into any site as a one-line iframe: its ?agent= and ?model= parameters, the hosts a model may come from, how a clip-less rig still comes to life, and the sentence the page shows for every way it can fail.",
+ "link": "/docs/avatar-artifact",
+ "tags": [
+ "launch"
+ ]
+ },
{
"date": "2026-08-16",
"type": "update",
@@ -133,6 +143,17 @@
"improvement"
]
},
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "Atlas: a long search term no longer knocks the page sideways, and the clear button is tappable",
+ "summary": "/atlas is the whole platform as one searchable map, and its search box had three rough edges. Searching for a very long unbroken term (a pasted URL, a run-on word) pushed the text past the edge of its box and gave the entire page a horizontal scrollbar, so the layout slid sideways under you while you typed. Long text now wraps wherever it has to, and the same rule keeps long route paths from stretching the result grid on a narrow phone. The box also drew two clear buttons side by side, the one the browser paints for a search field and the one the page draws, which every other search surface on the site already suppresses and this one had missed. The remaining clear button was a 16 by 18 pixel target, well under the minimum anyone should have to hit on a phone. It is now a 26 pixel square with its own hover and keyboard focus treatment.",
+ "link": "/atlas",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
{
"date": "2026-08-16",
"type": "update",
@@ -316,6 +337,17 @@
"fix"
]
},
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "Multi-LLM Brain now shows every model it can actually run, and says so when one is out",
+ "summary": "The /brain playground kept its own list of models on the page, and it had drifted from the models the server can really reach: nine were missing from the picker, including the current Claude Opus 5 and Sonnet 5 and all seven free models a signed-out visitor is allowed to use, while two entries in the list no longer existed at all. The picker is now built from the live model service, so it always matches what can be sent. A signed-out visitor opens on four models that are genuinely free to use instead of paid ones that could only refuse, and every model that needs an account is marked with a lock and a sign-in link rather than failing after you press Send. When the model list cannot be loaded the page now says so and offers a Retry that recovers, instead of quietly offering every model and erroring on send. When a model fails after every fallback route, the column now explains what happened and what to try next instead of printing a bare upstream word like Forbidden. Model choices are remembered between visits. On a phone the session list is reachable again through a drawer, the compare columns snap one per screen, and the Send button is no longer covered by the floating helper widgets. Model chips are proper buttons with keyboard focus, every field has a label, and clearing your persona from the sidebar now clears it everywhere.",
+ "link": "/brain",
+ "tags": [
+ "improvement",
+ "fix"
+ ]
+ },
{
"date": "2026-08-16",
"type": "update",
@@ -537,6 +569,17 @@
"improvement"
]
},
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "The Arena now ranks agents that actually traded above the ones that sat out",
+ "summary": "A tournament scored on realized P&L was handing a clean zero to every entrant that never opened a trade, which put them above every agent that took a real loss. On today's Daily Arena that meant the eight agents who did nothing held the whole podium while the two that traded sat at the bottom, and because a no-trade entrant can never clear the prize gates, the splits would have paid out nothing at all. Ranking now puts a result ahead of no result: entrants with no closed trade in the window sit below everyone who competed. They stay on the board with their honest zero, they just no longer outrank the field. The trophy, the podium on the front page and the prize ladder now only ever name an entrant that has actually traded, and the ladder only names an agent that is genuinely prize-eligible instead of promising a payout the settlement would never send. A tournament nobody has traded in now says it is awaiting the first ranked trade rather than crowning a zero. Also fixed: the tournament name in the browser tab was being reset to the generic page title a second after the page loaded, and the create-tournament form's fields overflowed their sheet on a 320px phone.",
+ "link": "/arena",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
{
"date": "2026-08-16",
"type": "update",
@@ -551,8 +594,42 @@
{
"date": "2026-08-16",
"type": "update",
- "title": "The Avatar CLI page now shows real command output, and its terminal is readable by a screen reader",
- "summary": "The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and one of its examples was staged rather than captured: the validate card showed a seven-error failure stitched together from errors no single manifest produces. It now shows a verbatim run against a genuinely broken manifest, with the instance paths (/, /mesh, /skeleton) the CLI really prints, so what you read on the page is what your shell will print back. The replayed terminal in the hero was marked up as one flat image, which meant a screen reader announced a one-line caption and hid both the transcript and the Replay button inside it; the transcript is now a named, readable region and Replay is an ordinary button again. The live viewer's failure notice keeps a link to the web component reference, and that link now looks like a link and shows a focus ring when you tab to it, as do the scrollable code blocks. The copy button's fallback hint used to tell everyone to press Command C; it now names the shortcut for the keyboard you are actually on.",
+ "title": "The ASL Alphabet drill now answers every press, and its package link goes somewhere real",
+ "summary": "Two things on /asl-alphabet did nothing when you used them. The Check button in the reading drill sat enabled from the moment the page loaded, but pressing it before starting a round, with an empty box, or after revealing an answer changed nothing on screen at all, which reads as a broken button. Check now always replies: it tells you to press start when there is no round to grade, tells you to type what you read when the box is empty, and says the round is finished once it has been graded or revealed. The start button's label follows the drill instead of being stuck on \"Skip to the next one\" after a round ends, so the feedback line can name the button you actually need. A wrong answer no longer says \"Watch it again\" when the round is already closed; it points at \"Show it again\" to study the shape and then move on. The drill's feedback is now announced to screen readers instead of changing silently. Separately, the note at the foot of the page (and the matching one on /sign-mirror) linked the signing engine to an npm page that does not exist: the package is in the repository and its registry publish has not landed yet, so both now link to the real source and say how to install it.",
+ "link": "/asl-alphabet",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "The assistant widget now tells you the truth when its chat lane is busy",
+ "summary": "Two things on the assistant widget page reported the wrong thing. The live preview status line always read \"Starting the live preview...\", even once the avatar was up and running, because the translation pass reset it a moment after the page wrote it. Every state now carries its own translated line, so the preview says what it is actually doing, and it reports a real failure with a Retry button when the widget or its avatar does not come up instead of silently claiming success. Second, when the free chat lane was throttled the widget only said \"Chat failed (HTTP 503)\", and a signed-out visitor could even be told to sign in for a capacity problem the widget has no sign-in for. The chat endpoint now answers with a proper at-capacity result and how long to wait, and the widget shows that in full and points at the settings panel where a visitor can drop in their own Groq or OpenRouter key. Editing the builder no longer re-downloads the avatar on every keystroke.",
+ "link": "/assistant",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "The Avatar Artifact page now shows real avatars, not just its house portrait",
+ "summary": "/avatar-artifact is the standalone viewer you can paste into any site as a bare iframe, but it only ever rendered one built-in procedural figure, so there was no way to point it at an actual avatar. It now takes ?agent=<id> to show any public agent's avatar, resolved live from the platform, or ?model=<glb url> to show a model directly from an allowed asset host. The agent's name becomes the heading and an Open profile link appears, the Copy embed button hands you an iframe of exactly what you are looking at, and avatars with no baked animation play the shared idle clip retargeted onto their own skeleton instead of standing in a T-pose. Every way it can fail now says what happened and what to do next: an id that matches no agent, an agent whose avatar is private or not created yet, a model URL from a host the viewer will not render, a model that would not load, and a browser without WebGL. Loading a large model reports its download progress instead of guessing, and a slow but healthy download is no longer cut off as a failure. With no parameters the page keeps its original portrait.",
+ "link": "/avatar-artifact",
+ "tags": [
+ "feature",
+ "improvement",
+ "fix"
+ ]
+ },
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "The Avatar CLI page terminal is readable by a screen reader, and its recovery link is reachable by keyboard",
+ "summary": "The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and the replayed terminal in its hero was marked up as one flat image. A screen reader announced a single caption for it and hid everything inside: the transcript itself, and the Replay button that restarts it. The transcript is now a named region a screen reader can read line by line, Replay is an ordinary button again, and the transcript reports itself as busy while it is still typing rather than being read half-written. Below it, the live viewer proves the embed snippet works, and when the component loader cannot be reached it swaps in a notice with a link to the web component reference. That link is the only way out of the failure, and it neither looked like a link nor showed a focus ring when tabbed to; it now does both. The scrollable code blocks, which the browser makes keyboard focusable, now show the site focus ring instead of a hairline default. And the copy button's clipboard fallback used to tell everyone to press Command C, including on Windows and Linux; it now names the shortcut for the keyboard you are actually on.",
"link": "/avatar-cli",
"tags": [
"fix",
@@ -581,6 +658,17 @@
"docs"
]
},
+ {
+ "date": "2026-08-16",
+ "type": "update",
+ "title": "The BNB block race stops showing a dead chain's last number as if it were live",
+ "summary": "The live block race compares BNB Chain, Base, Ethereum and Solana by measuring each one off a public RPC every few seconds, and it promises that a chain going quiet says so instead of freezing a fake number. The four lane cards kept that promise; the big headline number did not. When the BNB RPC stopped answering, the headline held its last reading at full brightness, kept claiming it was N times faster than Base, and the Updated stamp sat on \"just now\" no matter how long the outage ran. The headline now dims the moment the reading stops being live, labels it as last measured and prints its real age in seconds, and drops the speed comparison rather than race a stale number against a live one. The Updated stamp ages off the measurement time the server reports, and a poll that measured nothing no longer refreshes it. If every chain goes down at once the page now raises its outage notice and its Retry button, which previously could only ever appear before the first successful reading and so was unreachable during a real outage. Retry now shows it is working, makes a genuinely new request rather than reporting an older in-flight one's failure, and no longer sits underneath the floating helper widgets where the click was being swallowed. A lane that has never returned a sample says so plainly instead of pointing at a reading that is not on screen, each poll goes to the network instead of replaying a cached body up to fifteen seconds old, and screen readers now hear a lane announce a change of state rather than re-reading all four numbers twelve times a minute.",
+ "link": "/bnb-latency",
+ "tags": [
+ "fix",
+ "improvement"
+ ]
+ },
{
"date": "2026-08-16",
"type": "update",
diff --git a/public/changelog.xml b/public/changelog.xml
index 184be19ef..bde3a817e 100644
--- a/public/changelog.xml
+++ b/public/changelog.xml
@@ -6,6 +6,14 @@
<atom:link href="https://three.ws/changelog.xml" rel="self" type="application/rss+xml"/>
<description>What's new on three.ws: features, improvements, and releases, newest first.</description>
<language>en</language>
+ <item>
+ <title>Docs · Avatar Artifact viewer</title>
+ <link>https://three.ws/docs/avatar-artifact</link>
+ <guid isPermaLink="false">2026-08-16:Docs · Avatar Artifact viewer</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>launch</category>
+ <description>The standalone avatar viewer you paste into any site as a one-line iframe: its ?agent= and ?model= parameters, the hosts a model may come from, how a clip-less rig still comes to life, and the sentence the page shows for every way it can fail.</description>
+ </item>
<item>
<title>A coin world link now unfurls as the live world, and only ever points at three.ws</title>
<link>https://three.ws/play</link>
@@ -102,6 +110,14 @@
<category>fix</category>
<description>On the AR Studio page the site-wide floating language picker landed in the bottom-right corner, directly over the right-hand end of the control dock, so the Clear button (and the buttons beside it on a phone) could not be tapped at all. The picker now lives in the studio's own top bar, where it is always reachable and leaves the dock its full width. It also shrinks instead of pushing its neighbours off screen on narrow phones, which fixes the same crowding anywhere else it appears in a tight header. The page description now also names every place models can come from: your forges, the community feed, the free CC0 object library, and any GLB link.</description>
</item>
+ <item>
+ <title>Atlas: a long search term no longer knocks the page sideways, and the clear button is tappable</title>
+ <link>https://three.ws/atlas</link>
+ <guid isPermaLink="false">2026-08-16:Atlas: a long search term no longer knocks the page sideways, and the clear button is tappable</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>/atlas is the whole platform as one searchable map, and its search box had three rough edges. Searching for a very long unbroken term (a pasted URL, a run-on word) pushed the text past the edge of its box and gave the entire page a horizontal scrollbar, so the layout slid sideways under you while you typed. Long text now wraps wherever it has to, and the same rule keeps long route paths from stretching the result grid on a narrow phone. The box also drew two clear buttons side by side, the one the browser paints for a search field and the one the page draws, which every other search surface on the site already suppresses and this one had missed. The remaining clear button was a 16 by 18 pixel target, well under the minimum anyone should have to hit on a phone. It is now a 26 pixel square with its own hover and keyboard focus treatment.</description>
+ </item>
<item>
<title>Autopilot Activity keeps up when you switch agents, and its receipts link to the exact memory</title>
<link>https://three.ws/autopilot-activity</link>
@@ -238,6 +254,14 @@
<category>fix</category>
<description>The public revenue feed, the checkout SKU API and the ring report all took values straight from the URL into the database. A timestamp that was not a timestamp, a checkout id that was not an id, and the query fell over: the caller got an opaque &quot;something went wrong&quot; with a support reference, and our own alerting logged a fault that was only ever a typo. All three now say what was wrong, in words, with the right status code. The ring report also stops treating an unrecognised period as &quot;show me everything&quot;, which quietly answered the wrong question and ran the most expensive query it has. The revenue feed now echoes back the endpoint filter it actually applied rather than the raw text it was sent, and the paid-call lookup only accepts something shaped like a real transaction signature.</description>
</item>
+ <item>
+ <title>Multi-LLM Brain now shows every model it can actually run, and says so when one is out</title>
+ <link>https://three.ws/brain</link>
+ <guid isPermaLink="false">2026-08-16:Multi-LLM Brain now shows every model it can actually run, and says so when one is out</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>improvement</category>
+ <description>The /brain playground kept its own list of models on the page, and it had drifted from the models the server can really reach: nine were missing from the picker, including the current Claude Opus 5 and Sonnet 5 and all seven free models a signed-out visitor is allowed to use, while two entries in the list no longer existed at all. The picker is now built from the live model service, so it always matches what can be sent. A signed-out visitor opens on four models that are genuinely free to use instead of paid ones that could only refuse, and every model that needs an account is marked with a lock and a sign-in link rather than failing after you press Send. When the model list cannot be loaded the page now says so and offers a Retry that recovers, instead of quietly offering every model and erroring on send. When a model fails after every fallback route, the column now explains what happened and what to try next instead of printing a bare upstream word like Forbidden. Model choices are remembered between visits. On a phone the session list is reachable again through a drawer, the compare columns snap one per screen, and the Send button is no longer covered by the floating helper widgets. Model chips are proper buttons with keyboard focus, every field has a label, and clearing your persona from the sidebar now clears it everywhere.</description>
+ </item>
<item>
<title>Paying with $THREE works again: credits top-ups, tips and purchases stopped failing before they were signed</title>
<link>https://three.ws/credits</link>
@@ -406,6 +430,14 @@
<category>fix</category>
<description>The x402 Arbitrage page compares what different providers charge for the same capability, and it builds that comparison from several facilitator catalogs at once. When one of those catalogs was unreachable, its listings simply vanished from every comparison and the page said nothing about it. A scan that had only seen half the market looked exactly like a market with fewer price gaps in it, which is the opposite of what the page is for. Each scan now names the facilitators it reached and how many listings each one contributed, and calls out any that dropped out along with the reason, so a short board is never mistaken for a quiet one. Separately, closing the payment window without paying used to leave the Pay cheapest button dead for four seconds before it would accept another click. Deciding not to pay now hands the button straight back.</description>
</item>
+ <item>
+ <title>The Arena now ranks agents that actually traded above the ones that sat out</title>
+ <link>https://three.ws/arena</link>
+ <guid isPermaLink="false">2026-08-16:The Arena now ranks agents that actually traded above the ones that sat out</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>A tournament scored on realized P&amp;L was handing a clean zero to every entrant that never opened a trade, which put them above every agent that took a real loss. On today's Daily Arena that meant the eight agents who did nothing held the whole podium while the two that traded sat at the bottom, and because a no-trade entrant can never clear the prize gates, the splits would have paid out nothing at all. Ranking now puts a result ahead of no result: entrants with no closed trade in the window sit below everyone who competed. They stay on the board with their honest zero, they just no longer outrank the field. The trophy, the podium on the front page and the prize ladder now only ever name an entrant that has actually traded, and the ladder only names an agent that is genuinely prize-eligible instead of promising a payout the settlement would never send. A tournament nobody has traded in now says it is awaiting the first ranked trade rather than crowning a zero. Also fixed: the tournament name in the browser tab was being reset to the generic page title a second after the page loaded, and the create-tournament form's fields overflowed their sheet on a 320px phone.</description>
+ </item>
<item>
<title>The artifact builder now reports what it actually measures, and stops flagging legal artifacts as over budget</title>
<link>https://three.ws/artifact</link>
@@ -415,12 +447,36 @@
<description>The /artifact page is the pre-flight for pasting a three.ws agent into a Claude conversation: it builds the self-contained HTML, renders it in a sandbox that mirrors Claude's, and reports what you are about to paste. Two of those numbers were misleading. The stat labelled First paint was really the time the endpoint took to hand over the bytes; nothing had been rendered yet when it was measured. It is now split into Fetch, which is that download, and Render, which times how long the sandboxed frame takes to parse the document and start the viewer. The size gauge was grading the finished artifact against the size limit for the model inside it, but inlining base64-encodes the model and costs four bytes for every three, so a perfectly acceptable 7 MB artifact was painted amber and anything built from a model over about 5.6 MB went red, contradicting the endpoint that had just accepted it. Size is now graded against the roughly 8.6 MB an accepted model can actually produce, and hovering the number tells you how much of that budget you have used. The Size budget note explained the same thing incorrectly and now spells out the base64 cost.</description>
</item>
<item>
- <title>The Avatar CLI page now shows real command output, and its terminal is readable by a screen reader</title>
+ <title>The ASL Alphabet drill now answers every press, and its package link goes somewhere real</title>
+ <link>https://three.ws/asl-alphabet</link>
+ <guid isPermaLink="false">2026-08-16:The ASL Alphabet drill now answers every press, and its package link goes somewhere real</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>Two things on /asl-alphabet did nothing when you used them. The Check button in the reading drill sat enabled from the moment the page loaded, but pressing it before starting a round, with an empty box, or after revealing an answer changed nothing on screen at all, which reads as a broken button. Check now always replies: it tells you to press start when there is no round to grade, tells you to type what you read when the box is empty, and says the round is finished once it has been graded or revealed. The start button's label follows the drill instead of being stuck on &quot;Skip to the next one&quot; after a round ends, so the feedback line can name the button you actually need. A wrong answer no longer says &quot;Watch it again&quot; when the round is already closed; it points at &quot;Show it again&quot; to study the shape and then move on. The drill's feedback is now announced to screen readers instead of changing silently. Separately, the note at the foot of the page (and the matching one on /sign-mirror) linked the signing engine to an npm page that does not exist: the package is in the repository and its registry publish has not landed yet, so both now link to the real source and say how to install it.</description>
+ </item>
+ <item>
+ <title>The assistant widget now tells you the truth when its chat lane is busy</title>
+ <link>https://three.ws/assistant</link>
+ <guid isPermaLink="false">2026-08-16:The assistant widget now tells you the truth when its chat lane is busy</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>Two things on the assistant widget page reported the wrong thing. The live preview status line always read &quot;Starting the live preview...&quot;, even once the avatar was up and running, because the translation pass reset it a moment after the page wrote it. Every state now carries its own translated line, so the preview says what it is actually doing, and it reports a real failure with a Retry button when the widget or its avatar does not come up instead of silently claiming success. Second, when the free chat lane was throttled the widget only said &quot;Chat failed (HTTP 503)&quot;, and a signed-out visitor could even be told to sign in for a capacity problem the widget has no sign-in for. The chat endpoint now answers with a proper at-capacity result and how long to wait, and the widget shows that in full and points at the settings panel where a visitor can drop in their own Groq or OpenRouter key. Editing the builder no longer re-downloads the avatar on every keystroke.</description>
+ </item>
+ <item>
+ <title>The Avatar Artifact page now shows real avatars, not just its house portrait</title>
+ <link>https://three.ws/avatar-artifact</link>
+ <guid isPermaLink="false">2026-08-16:The Avatar Artifact page now shows real avatars, not just its house portrait</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>feature</category>
+ <description>/avatar-artifact is the standalone viewer you can paste into any site as a bare iframe, but it only ever rendered one built-in procedural figure, so there was no way to point it at an actual avatar. It now takes ?agent=&lt;id&gt; to show any public agent's avatar, resolved live from the platform, or ?model=&lt;glb url&gt; to show a model directly from an allowed asset host. The agent's name becomes the heading and an Open profile link appears, the Copy embed button hands you an iframe of exactly what you are looking at, and avatars with no baked animation play the shared idle clip retargeted onto their own skeleton instead of standing in a T-pose. Every way it can fail now says what happened and what to do next: an id that matches no agent, an agent whose avatar is private or not created yet, a model URL from a host the viewer will not render, a model that would not load, and a browser without WebGL. Loading a large model reports its download progress instead of guessing, and a slow but healthy download is no longer cut off as a failure. With no parameters the page keeps its original portrait.</description>
+ </item>
+ <item>
+ <title>The Avatar CLI page terminal is readable by a screen reader, and its recovery link is reachable by keyboard</title>
<link>https://three.ws/avatar-cli</link>
- <guid isPermaLink="false">2026-08-16:The Avatar CLI page now shows real command output, and its terminal is readable by a screen reader</guid>
+ <guid isPermaLink="false">2026-08-16:The Avatar CLI page terminal is readable by a screen reader, and its recovery link is reachable by keyboard</guid>
<pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
<category>fix</category>
- <description>The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and one of its examples was staged rather than captured: the validate card showed a seven-error failure stitched together from errors no single manifest produces. It now shows a verbatim run against a genuinely broken manifest, with the instance paths (/, /mesh, /skeleton) the CLI really prints, so what you read on the page is what your shell will print back. The replayed terminal in the hero was marked up as one flat image, which meant a screen reader announced a one-line caption and hid both the transcript and the Replay button inside it; the transcript is now a named, readable region and Replay is an ordinary button again. The live viewer's failure notice keeps a link to the web component reference, and that link now looks like a link and shows a focus ring when you tab to it, as do the scrollable code blocks. The copy button's fallback hint used to tell everyone to press Command C; it now names the shortcut for the keyboard you are actually on.</description>
+ <description>The /avatar-cli page teaches the four commands of @three-ws/avatar-cli, and the replayed terminal in its hero was marked up as one flat image. A screen reader announced a single caption for it and hid everything inside: the transcript itself, and the Replay button that restarts it. The transcript is now a named region a screen reader can read line by line, Replay is an ordinary button again, and the transcript reports itself as busy while it is still typing rather than being read half-written. Below it, the live viewer proves the embed snippet works, and when the component loader cannot be reached it swaps in a notice with a link to the web component reference. That link is the only way out of the failure, and it neither looked like a link nor showed a focus ring when tabbed to; it now does both. The scrollable code blocks, which the browser makes keyboard focusable, now show the site focus ring instead of a hairline default. And the copy button's clipboard fallback used to tell everyone to press Command C, including on Windows and Linux; it now names the shortcut for the keyboard you are actually on.</description>
</item>
<item>
<title>The AWS page stops promising AWS invoicing three.ws does not do, and the GLB optimizer reads compressed avatars again</title>
@@ -438,6 +494,14 @@
<category>improvement</category>
<description>Every avatar that finishes a scan or a prompt is given a draft on-chain identity in the background. That step is deliberately best-effort, so it can never cost you your avatar, but it was also completely silent: whether it minted, was skipped because the deployment holds no minting key, or was switched off entirely, the logs looked identical, which is how a run of avatars can go by without anyone noticing the identity was never created. Each finished avatar now records the outcome of both chains, including the reason a leg was skipped. The documentation for the feature also gained a way to prove the whole path on a machine that cannot reach the public test-network faucet, which is routinely exhausted on shared build machines: run your own node with the real Metaplex programs, and the proof reports which node served it rather than handing back an explorer link that leads nowhere.</description>
</item>
+ <item>
+ <title>The BNB block race stops showing a dead chain's last number as if it were live</title>
+ <link>https://three.ws/bnb-latency</link>
+ <guid isPermaLink="false">2026-08-16:The BNB block race stops showing a dead chain's last number as if it were live</guid>
+ <pubDate>Sun, 16 Aug 2026 12:00:00 GMT</pubDate>
+ <category>fix</category>
+ <description>The live block race compares BNB Chain, Base, Ethereum and Solana by measuring each one off a public RPC every few seconds, and it promises that a chain going quiet says so instead of freezing a fake number. The four lane cards kept that promise; the big headline number did not. When the BNB RPC stopped answering, the headline held its last reading at full brightness, kept claiming it was N times faster than Base, and the Updated stamp sat on &quot;just now&quot; no matter how long the outage ran. The headline now dims the moment the reading stops being live, labels it as last measured and prints its real age in seconds, and drops the speed comparison rather than race a stale number against a live one. The Updated stamp ages off the measurement time the server reports, and a poll that measured nothing no longer refreshes it. If every chain goes down at once the page now raises its outage notice and its Retry button, which previously could only ever appear before the first successful reading and so was unreachable during a real outage. Retry now shows it is working, makes a genuinely new request rather than reporting an older in-flight one's failure, and no longer sits underneath the floating helper widgets where the click was being swallowed. A lane that has never returned a sample says so plainly instead of pointing at a reading that is not on screen, each poll goes to the network instead of replaying a cached body up to fifteen seconds old, and screen readers now hear a lane announce a change of state rather than re-reading all four numbers twelve times a minute.</description>
+ </item>
<item>
<title>The BNB Chain page now says how many demos are actually live</title>
<link>https://three.ws/bnb</link>