lm MCP

v3.2.3

io.github.houtini-ai/lm

Connect Claude to any OpenAI-compatible LLM endpoint and offload routine work to a local model.

transport:stdio runtime:npm

Target client

run in your project directory

claude mcp add lm -e HOUTINI_LM_ENDPOINT_URL=<HOUTINI_LM_ENDPOINT_URL> -e HOUTINI_LM_API_KEY=<HOUTINI_LM_API_KEY> -e HOUTINI_LM_MODEL=<HOUTINI_LM_MODEL> -e HOUTINI_LM_PROVIDER=<HOUTINI_LM_PROVIDER> -- npx -y @houtini/lm

Adds it for this project only. Append --scope user to make it available everywhere. Claude Code docs

This listing does not declare its tools. Connect the server and your client will discover them on the handshake.

Secrets and scopes

The registry records which variables this server needs, not what privileges they must carry. Issue each one with the narrowest scope that works.

HOUTINI_LM_ENDPOINT_URL
HOUTINI_LM_API_KEY
HOUTINI_LM_MODEL
HOUTINI_LM_PROVIDER

Runtime

Transport stdio

Registry npm

Package @houtini/lm

Version 3.2.3

Provenance Official registry

Package managers

npx npm pnpm yarn bun Homebrew apt Chocolatey winget conda

npx -y @houtini/lm

npm install -g @houtini/lm

pnpm dlx @houtini/lm

yarn dlx @houtini/lm

bunx @houtini/lm

Want a slot? Make an offer

Registry record

Repository https://github.com/houtini-ai/lm

Synced 15 h ago

JSON API /api/v0/servers/io.github.houtini-ai/lm