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(mcp): resync the golden contracts and public catalog with the tool text that already shipped

nirholas8d54b93688df
diff --git a/prompts/swarm-100/sweep-gate.md b/prompts/swarm-100/sweep-gate.md
deleted file mode 100644
index ea0ac8928..000000000
--- a/prompts/swarm-100/sweep-gate.md
+++ /dev/null
@@ -1,86 +0,0 @@
-# Sweep: Repo gate green
-
-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
-
-Run the sweep's own command first and work from its output, not from any
-memory of what used to fail:
-
- npm run gate
-
-Capture the full output to the session scratchpad; do not work from an
-excerpt.
-
-## Task
-
-Run the full repo gate. Fix every failing stage at its root cause, in order of appearance; never mask a check or skip a stage. The gate chains conflict, claude-md, cron, MCP, route, handler, page, docs, tutorial, guard, token, and tour audits; each failure names its own fix.
-
-Fix everything found in this sweep's scope. Findings that belong to a
-different surface entirely (and would take this session off its one task) go
-in the report as named follow-ups instead; everything in scope gets fixed
-here, at root, with no masking.
-
-## Definition of done
-
-- [ ] The command above exits 0 (or, for measurement-style sweeps, the report
- carries the measured numbers and every committed fix).
-- [ ] Every fix is a root-cause fix; no check was weakened, skipped, or
- quarantined to get to green.
-- [ ] npm run check:rules -- --paths <touched files> exits 0; npm test passes
- if code changed.
-
-## 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/sweep-gate.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/mcp-catalog.json b/public/mcp-catalog.json
index 0c48ce6d9..7d4e739cc 100644
--- a/public/mcp-catalog.json
+++ b/public/mcp-catalog.json
@@ -4463,7 +4463,7 @@
{
"name": "chat",
"title": "Run a chat completion",
- "description": "Run a chat completion through the three.ws LLM router and get back the full reply as a single message. Pick `provider` from `list_providers` (default `gpt-oss-120b`, a free open-weight model). `messages` is the conversation so far (roles `user`/`assistant`); add system instructions via `system`. The route streams and auto-falls-back across mirrors/free tiers if the chosen model is briefly down — this tool collapses that into one reply and reports the actual route taken, token usage, and timing. Paid first-party flagships (Claude, GPT-4o, o3, Qwen Plus, DeepSeek) need a three.ws API key (set THREE_WS_API_KEY); the free open-weight tiers (GPT-OSS 120B, NVIDIA NIM models) need none. Read-only: a completion does not change any platform state.",
+ "description": "Run a chat completion through the three.ws LLM router and get back the full reply as a single message. Pick `provider` from `list_providers` (default `gpt-oss-120b`, a free open-weight model). `messages` is the conversation so far (roles `user`/`assistant`); add system instructions via `system`. The route streams and auto-falls-back across mirrors/free tiers if the chosen model is briefly down — this tool collapses that into one reply and reports the actual route taken, token usage, and timing. Paid first-party flagships (Claude, GPT-5, o3, Grok, Qwen Plus, DeepSeek) need a three.ws API key (set THREE_WS_API_KEY); the free open-weight tiers (GPT-OSS 120B, NVIDIA NIM models) need none. Read-only: a completion does not change any platform state.",
"server": {
"id": "brain-mcp",
"title": "@three-ws/brain-mcp",
@@ -4518,7 +4518,7 @@
"provider": {
"type": "string",
"minLength": 1,
- "description": "Provider/model key from list_providers (e.g. \"claude-sonnet-4-6\", \"gpt-4o\", \"deepseek-r1\"). Defaults to \"gpt-oss-120b\"."
+ "description": "Provider/model key from list_providers (e.g. \"claude-sonnet-5\", \"gpt-5.4\", \"deepseek-r1\"). Defaults to \"gpt-oss-120b\"."
},
"system": {
"type": "string",
@@ -4542,7 +4542,7 @@
{
"name": "list_providers",
"title": "List LLM providers and models",
- "description": "List every LLM the three.ws router can run a completion through — one entry per provider/model. Each entry has: `key` (the id you pass to `chat`), `label` (human name), `network` (e.g. Anthropic, OpenAI, DashScope, NVIDIA NIM, IBM watsonx.ai), `tier` (flagship | balanced | fast | reasoning), `maxOutput` (max output tokens), `description`, `available` (true when the server currently holds a working key/route for it), and `requiresAuth` (true when it is a paid first-party model that needs a three.ws API key — set THREE_WS_API_KEY; false for the free open-weight tiers). Call this first to pick a `key` for `chat`. The set is live, so availability moves between calls — not idempotent.",
+ "description": "List every LLM the three.ws router can run a completion through — one entry per provider/model. Each entry has: `key` (the id you pass to `chat`), `label` (human name), `network` (e.g. Anthropic, OpenAI, xAI, DashScope, NVIDIA NIM, IBM watsonx.ai), `tier` (flagship | balanced | fast | reasoning | pro | coding), `maxOutput` (max output tokens), `description`, `available` (true when the server currently holds a working key/route for it), and `requiresAuth` (true when it is a paid first-party model that needs a three.ws API key — set THREE_WS_API_KEY; false for the free open-weight tiers). Call this first to pick a `key` for `chat`. The set is live, so availability moves between calls — not idempotent.",
"server": {
"id": "brain-mcp",
"title": "@three-ws/brain-mcp",
@@ -7440,7 +7440,7 @@
{
"name": "build_world",
"title": "Build a complete 3D world from one sentence (no browser needed)",
- "description": "One sentence → a fully forged, exported 3D world, entirely server-side: composes the scene plan, forges every object on the free text→3D lane, and merges the result into one GLB with named, selectable nodes plus ground and lighting. Returns the populated diorama, the merged GLB URL, and a ready-to-open Scene Studio link. This is the whole progressive /diorama browser flow collapsed into one call — it can take a couple of minutes for a full object set. Partial forges are not failures: whatever forged is exported, and `skipped` names anything that did not.",
+ "description": "One sentence → a fully forged, exported 3D world, entirely server-side: composes the scene plan, forges every object on the free text→3D lane, and merges the result into one GLB with named, selectable nodes plus ground and lighting. Returns the populated diorama, the merged GLB URL, and a ready-to-open Scene Studio link. This is the whole progressive /diorama browser flow collapsed into one call; it can take a couple of minutes for a full object set. Partial forges are not failures: whatever forged is exported, and `skipped` names anything that did not.",
"server": {
"id": "scene-mcp",
"title": "@three-ws/scene-mcp",
@@ -7479,7 +7479,7 @@
{
"name": "compose_scene",
"title": "Compose a 3D diorama plan from one sentence",
- "description": "Compose a tiny 3D diorama from one short sentence. Returns a PLAN: an evocative title, a mood (dawn/day/dusk/night), ground + island type, a color palette, and 2–8 placed objects — each with a single-object forge prompt, position, scale and rotation. No meshes are generated yet and nothing is saved; feed the plan to the three.ws diorama page (or /api/forge per object) to build the world. Read-mostly but runs live LLM inference.",
+ "description": "Compose a tiny 3D diorama from one short sentence. Returns a PLAN: an evocative title, a mood (dawn/day/dusk/night), ground + island type, a color palette, and 2-8 placed objects, each with a single-object forge prompt, position, scale and rotation. No meshes are generated yet and nothing is saved; feed the plan to the three.ws diorama page (or /api/forge per object) to build the world. Read-mostly but runs live LLM inference.",
"server": {
"id": "scene-mcp",
"title": "@three-ws/scene-mcp",
@@ -7518,7 +7518,7 @@
{
"name": "export_scene",
"title": "Export a forged diorama as one GLB scene",
- "description": "Merge an already-forged diorama (objects with real glbUrl values, e.g. from your own forging after compose_scene, or fetched via get_scene) into ONE glTF 2.0 binary: every object becomes a named, selectable node, plus a real ground disc and mood-tuned lighting. Returns the merged GLB URL and a ready-to-open three.ws Scene Studio link. Objects that never forged are skipped and reported, not treated as a failure — a partial world still exports. Requires the target deployment to have object storage configured (three.ws does).",
+ "description": "Merge an already-forged diorama (objects with real glbUrl values, e.g. from your own forging after compose_scene, or fetched via get_scene) into ONE glTF 2.0 binary: every object becomes a named, selectable node, plus a real ground disc and mood-tuned lighting. Returns the merged GLB URL and a ready-to-open three.ws Scene Studio link. Objects that never forged are skipped and reported rather than treated as a failure; a partial world still exports. Requires the target deployment to have object storage configured (three.ws does).",
"server": {
"id": "scene-mcp",
"title": "@three-ws/scene-mcp",
@@ -7543,7 +7543,7 @@
"diorama": {
"type": "object",
"additionalProperties": {},
- "description": "The diorama object to export — same shape returned by compose_scene/get_scene, with each object you want included carrying status:\"ready\" and a real glbUrl."
+ "description": "The diorama object to export, same shape returned by compose_scene/get_scene, with each object you want included carrying status:\"ready\" and a real glbUrl."
}
},
"required": [
@@ -7594,7 +7594,7 @@
{
"name": "list_scenes",
"title": "List recent or featured dioramas",
- "description": "Browse the public diorama gallery — the most recent saved worlds, or the featured set. Returns a list of cards (id, title, mood, thumbnail/preview, view count) plus each world's viewer URL. Read-only.",
+ "description": "Browse the public diorama gallery: the most recent saved worlds, or the featured set. Returns a list of cards (id, title, mood, thumbnail/preview, view count) plus each world's viewer URL. Read-only.",
"server": {
"id": "scene-mcp",
"title": "@three-ws/scene-mcp",
@@ -7629,7 +7629,7 @@
"type": "integer",
"minimum": 1,
"maximum": 50,
- "description": "How many to return (1–50, default 24)."
+ "description": "How many to return (1-50, default 24)."
}
},
"additionalProperties": false
diff --git a/tests/fixtures/mcp-golden-tools.json b/tests/fixtures/mcp-golden-tools.json
index f500f7693..b9077d187 100644
--- a/tests/fixtures/mcp-golden-tools.json
+++ b/tests/fixtures/mcp-golden-tools.json
@@ -2077,7 +2077,7 @@
"idempotentHint": false,
"openWorldHint": true
},
- "schemaHash": "6abff79eeaf6"
+ "schemaHash": "0fd23416411d"
}
],
"packages/brain-mcp/src/tools/list-providers.js": [
@@ -2787,7 +2787,7 @@
"name": "build_world",
"title": "Build a complete 3D world from one sentence (no browser needed)",
"descHead": "One sentence → a fully forged, exported 3D world, entirely server-side: ",
- "descHash": "2ddf7bfcb02d",
+ "descHash": "0216a43f7396",
"annotations": {
"readOnlyHint": false,
"destructiveHint": false,
@@ -2802,7 +2802,7 @@
"name": "compose_scene",
"title": "Compose a 3D diorama plan from one sentence",
"descHead": "Compose a tiny 3D diorama from one short sentence. Returns a PLAN: an ev",
- "descHash": "d42bccef405d",
+ "descHash": "16e0ce6f8327",
"annotations": {
"readOnlyHint": false,
"destructiveHint": false,
@@ -2817,14 +2817,14 @@
"name": "export_scene",
"title": "Export a forged diorama as one GLB scene",
"descHead": "Merge an already-forged diorama (objects with real glbUrl values, e.g. f",
- "descHash": "910a519b23de",
+ "descHash": "ddccc838e819",
"annotations": {
"readOnlyHint": false,
"destructiveHint": false,
"idempotentHint": false,
"openWorldHint": true
},
- "schemaHash": "ac725edfea3d"
+ "schemaHash": "a0addfff5cc7"
}
],
"packages/scene-mcp/src/tools/get-scene.js": [
@@ -2845,14 +2845,14 @@
{
"name": "list_scenes",
"title": "List recent or featured dioramas",
- "descHead": "Browse the public diorama gallery — the most recent saved worlds, or the",
- "descHash": "79e7f4cc0eb6",
+ "descHead": "Browse the public diorama gallery: the most recent saved worlds, or the ",
+ "descHash": "13f98cc2aaa2",
"annotations": {
"readOnlyHint": true,
"idempotentHint": false,
"openWorldHint": true
},
- "schemaHash": "2b69c8fade89"
+ "schemaHash": "ccec143f22b2"
}
],
"packages/signals-mcp/src/tools/get-mirror-leaderboard.js": [