What problem does it solve? AI agents cannot call your Litestar API endpoints out of the box. This Skill guides you through exposing existing Litestar route handlers as Model Context Protocol (MCP) tools and resources over JSON-RPC 2.0, so MCP clients can discover and invoke your API without writing a separate server. ## Core Features & Use Cases - Automatic route discovery: GET handlers become MCP resources and POST/PUT/PATCH/DELETE handlers become MCP tools, discovered via the Litestar OpenAPI schema. - Per-route control: Override names and types with @mcp_tool, @mcp_resource, or opt dicts, and filter exposure with include/exclude tags and operations. - Auth and validation: Protect the MCP endpoint with OAuth 2.1 (PKCE) or Litestar guards, with tool arguments validated against OpenAPI request schemas. - Use Case: You have an e-commerce Litestar API and want an AI assistant to list products and add items to carts. Add LitestarMCP with include_tags=["public"], and the agent can call tools/list and tools/call at POST /mcp/ while internal metrics routes stay hidden. ## Quick Start Ask the AI to add the LitestarMCP plugin with an MCPConfig to your Litestar app and expose only your public-tagged routes as MCP tools and resources.