MCP client integration

Start the server over stdio, then point your client at it.

pydocs-mcp serve /path/to/project

Claude Code (~/.config/claude-code/mcp_servers.json or workspace .claude/mcp_servers.json):

{
  "mcpServers": {
    "pydocs": { "command": "pydocs-mcp", "args": ["serve", "/path/to/your/project"] }
  }
}

Cursor (~/.cursor/mcp.json or .cursor/mcp.json):

{
  "mcpServers": [
    { "name": "pydocs", "command": "pydocs-mcp", "args": ["serve", "/path/to/your/project"] }
  ]
}

Continue.dev (~/.continue/config.json):

{
  "mcpServers": [
    { "name": "pydocs", "command": "pydocs-mcp", "args": ["serve", "/path/to/your/project"] }
  ]
}

Example client invocations (ask your LLM to run these once connected):

search_codebase("batch inference vllm", kind="api", package="vllm", limit=20)  # limit caps multi-repo unions only
get_symbol("fastapi.routing.APIRouter")
get_references("fastapi.routing.APIRouter.include_router", direction="callers")
get_references("requests.auth.HTTPBasicAuth", direction="inherits")