Loom — the three.ws community 3D-creation gallery, from any AI agent. Browse the public feed of community-forged GLB creations, fetch one with its viewer URL, and submit your own.
A Model Context Protocol server that exposes Loom, the three.ws community 3D-creation gallery, over stdio. Browse what others have forged, pull any creation with a ready-to-embed viewer URL, and contribute your own GLB to the public feed.
Loom closes the loop with the rest of the three.ws 3D suite: forge a model with @three-ws/avatar-agent or compose a world with @three-ws/scene-mcp, then publish it here for the whole platform to see. Every read is live and public; submitting needs no key or account — it is gated server-side only by an IP rate limit.
npm install @three-ws/loom-mcpOr run with npx (no install):
npx @three-ws/loom-mcpClaude Code, one line:
claude mcp add loom -- npx -y @three-ws/loom-mcpClaude Desktop / Cursor (claude_desktop_config.json or mcp.json):
{
"mcpServers": {
"loom": {
"command": "npx",
"args": ["-y", "@three-ws/loom-mcp"]
}
}
}Inspect the surface with the MCP Inspector:
npx -y @modelcontextprotocol/inspector npx @three-ws/loom-mcp| Tool | Type | What it does |
|---|---|---|
get_loom_feed |
read-only | Browse the public gallery of community-forged 3D creations, newest-first, paginated via a nextBefore cursor. |
get_creation |
read-only | Fetch one creation by id, with its full metadata and an inline-previewable viewer URL + iframe. |
submit_creation |
write | Contribute a forged GLB to the public gallery. No key required; rate-limited (~20/hour per IP). |
Each read returns the creation's glbUrl plus a viewer_url (the /forge/embed orbit + AR viewer), an og_image_url (social card), and a paste-ready iframe_snippet, so any MCP client can preview the model inline.
get_loom_feed — limit (1–120, default 60), before (ms-epoch cursor; pass the prior page's nextBefore to page backwards).
get_creation — id (required — the creation UUID).
submit_creation — prompt (required, ≤1000 chars), glbUrl (required — see allowed hosts below), author (optional attribution, ≤40 chars, defaults to anon), previewImageUrl, tier, backend (all optional).
submit_creation posts to the public, world-readable gallery — only submit models you intend to share. There is no account, key, or signer: the endpoint is open and gated server-side by an IP rate limit (~20 submissions/hour) plus strict input sanitization. author is free-text attribution, not an authenticated identity.
The glbUrl must be an https URL hosted on one of these domains (enforced both client-side here and server-side):
three.ws · r2.dev · cloudflarestorage.com · replicate.delivery · githubusercontent.com
Re-submitting the exact same glbUrl returns the existing creation instead of duplicating it. The dedupe is best-effort by design: it scans only the newest few entries, which is where an accidental re-POST lands. Re-submitting a GLB that has since scrolled down the feed creates a second record.
get_creation answers an id that no longer exists with a typed not_found error rather than an empty creation, so a stale id in a saved reference is unambiguous.
// get_loom_feed > { "limit": 2 } { "ok": true, "count": 2, "nextBefore": 1749999000000, "has_more": true, "creations": [ { "id": "…", "prompt": "…", "glbUrl": "…", "viewer_url": "…", "iframe_snippet": "…" } ] }
Drop a creation's iframe_snippet onto any web page, or open its viewer_url to orbit the model (with AR on supported devices).
Runnable examples live in examples/:
node examples/list-tools.mjs # every tool with its full input schema
node examples/browse-loom.mjs # a page of the gallery + one creation's embed, liveBoth spawn this server over stdio and read the live public gallery. They are
deliberately read-only: submit_creation posts to a world-readable feed, so no
example writes on your behalf. See examples/README.md
for expected output.
- Node.js >= 20.
- Network access to
https://three.ws(or your ownTHREE_WS_BASE).
| Variable | Required | Default |
|---|---|---|
THREE_WS_BASE |
no | https://three.ws |
THREE_WS_TIMEOUT_MS |
no | 20000 |
No API key, signer, or account is needed for any tool.
- Homepage: https://three.ws
- Forge gallery: https://three.ws/forge
- Changelog: https://three.ws/changelog
- Issues: https://github.com/nirholas/three.ws/issues
- License: Apache-2.0, see LICENSE
Part of the three.ws SDK suite — 3D AI agents, on-chain identity, and agent payments.
Website · Changelog · GitHub
If loom-mcp saves you time, star it on GitHub. Stars are how other developers and AI agents find the repositories worth trusting, and they cost you one click.
Know someone who would use it? Post on X · Share on Bluesky · Share on LinkedIn · Submit to Hacker News · Share on Reddit
Coding agents and LLM tooling can read this repo directly: AGENTS.md, llms.txt, llms-full.txt. Point an agent at https://github.com/nirholas/loom-mcp and it has the context it needs.
- All repositories by nirholas: the full catalog, grouped by topic
- three.ws: the platform for 3D AI agents with Solana wallets, a skill marketplace and x402 payments
- Questions or ideas: open an issue or start a discussion
three.ws is a platform for 3D AI agents with Solana wallets: avatars, a skill marketplace, x402 payments and more than seventy MCP servers. @three-ws/loom-mcp is one package from it.
- Documentation site: https://nirholas.github.io/loom-mcp/
- npm: https://www.npmjs.com/package/@three-ws/loom-mcp
- Canonical source: https://github.com/nirholas/three.ws (this repository is a generated mirror of
packages/loom-mcp) - Agent-readable summary: llms.txt and AGENTS.md
- Issues and ideas: https://github.com/nirholas/loom-mcp/issues