lm for VS Code

io.github.houtini-ai/lm

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

client:VS Code transport:stdio runtime:npm

Install lm in VS Code

.vscode/mcp.json

{
  "inputs": [
    {
      "type": "promptString",
      "id": "houtini-lm-endpoint-url",
      "description": "HOUTINI_LM_ENDPOINT_URL",
      "password": true
    },
    {
      "type": "promptString",
      "id": "houtini-lm-api-key",
      "description": "HOUTINI_LM_API_KEY",
      "password": true
    },
    {
      "type": "promptString",
      "id": "houtini-lm-model",
      "description": "HOUTINI_LM_MODEL",
      "password": true
    },
    {
      "type": "promptString",
      "id": "houtini-lm-provider",
      "description": "HOUTINI_LM_PROVIDER",
      "password": true
    }
  ],
  "servers": {
    "lm": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@houtini/lm"
      ],
      "env": {
        "HOUTINI_LM_ENDPOINT_URL": "${input:houtini-lm-endpoint-url}",
        "HOUTINI_LM_API_KEY": "${input:houtini-lm-api-key}",
        "HOUTINI_LM_MODEL": "${input:houtini-lm-model}",
        "HOUTINI_LM_PROVIDER": "${input:houtini-lm-provider}"
      }
    }
  }
}

VS Code prompts for each input the first time the server starts, so no secret is written to the file. VS Code docs

lm in other clients

lm MCP server