Three Ws
by nirholascommunityv1.0.1
74
19
three.ws is a browser-based platform for rendering GLB/glTF 3D models, attaching an LLM-powered agent to them, optionally registering the agent on-chain as an ERC-8004 identity, and embedding the result as a web component or widget. It also exposes an MCP endpoint so external AI systems can control these agents programmatically.
From the registry: 3D avatars, embeds, glTF tools, agent memory, and on-chain agent identity from three.ws.
Install & configure
Prerequisites
- optionalDeploy or access a running three.ws serverThe MCP server is exposed by the three.ws backend as an HTTP endpoint (`api/mcp.js`). You need a reachable hosted instance or a local deployment of the full platform.
https://three.ws/$ git clone https://github.com/nirholas/3D-Agent.git && cd 3D-Agent && npm install && cp .env.example .env.local && psql $DATABASE_URL < api/_lib/schema.sql && npm run dev - optionalInstall Node.js 20+ and npm 10+Required to run the full platform locally.
- optionalProvision Postgres databaseA Neon Postgres database or any Postgres 15+ instance is required for local deployment.
- optionalSet required backend environmentAt minimum for local deployment: `PUBLIC_APP_ORIGIN`, `DATABASE_URL`, `JWT_SECRET`, and `ANTHROPIC_API_KEY`. Additional services like Cloudflare R2 may also be needed depending on features used.
$ cp .env.example .env.local
Install in your client
Paste into the AI you're chatting with — it will figure out the rest
Please install the `three-ws` MCP server into my current AI client (that's you).
Optional prerequisites:
- Deploy or access a running three.ws server — The MCP server is exposed by the three.ws backend as an HTTP endpoint (`api/mcp.js`). You need a reachable hosted instance or a local deployment of the full platform. Run: `git clone https://github.com/nirholas/3D-Agent.git && cd 3D-Agent && npm install && cp .env.example .env.local && psql $DATABASE_URL < api/_lib/schema.sql && npm run dev` (https://three.ws/)
- Install Node.js 20+ and npm 10+ — Required to run the full platform locally.
- Provision Postgres database — A Neon Postgres database or any Postgres 15+ instance is required for local deployment.
- Set required backend environment — At minimum for local deployment: `PUBLIC_APP_ORIGIN`, `DATABASE_URL`, `JWT_SECRET`, and `ANTHROPIC_API_KEY`. Additional services like Cloudflare R2 may also be needed depending on features used. Run: `cp .env.example .env.local`
Canonical MCP server config (HTTP transport):
- url: `<your-mcp-server-url>`
Note: README states the project exposes 'MCP (Model Context Protocol) over HTTP with JSON-RPC 2.0'. No client-facing MCP auth header format, fixed endpoint path, or standalone MCP server launch command is documented in the provided README. This appears to be part of the full three.ws backend rather than a separate stdio package.
Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.Note: README states the project exposes 'MCP (Model Context Protocol) over HTTP with JSON-RPC 2.0'. No client-facing MCP auth header format, fixed endpoint path, or standalone MCP server launch command is documented in the provided README. This appears to be part of the full three.ws backend rather than a separate stdio package.
Tools (6)
wave
Makes the agent wave
lookAt
Moves the agent to look at a target
play_clip
Plays an animation or motion clip
setExpression
Sets the agent's facial expression
speak
Makes the agent speak
remember
Stores information in the agent's memory
Prerequisites
- • Node.js 20+
Remote Endpoints
https://three.ws/api/mcp
type: streamable-http
Details
- Language
- JavaScript
- License
- NOASSERTION
- Released
- Jun 12, 2026
- Last commit
- Jun 29, 2026
- Registry name
- io.github.nirholas/three.ws