MCP server

Connect your coding agent to the Lauren docs

Lauren runs a remote Model Context Protocol server. Point Claude Code, Cursor, VS Code, Windsurf, Gemini CLI or any MCP client at it, and your agent searches and reads the current documentation instead of guessing from stale training data.

MCPhttps://lauren-py.dev/api/mcp

Streamable HTTPRead-onlyNo API keyFree

Add it to your client

One server name — lauren-docs — and one URL. Pick your tool and paste.

Run this once in your terminal:

Terminal

claude mcp add --transport http lauren-docs https://lauren-py.dev/api/mcp

.mcp.json — project scope, shared with your team

{
  "mcpServers": {
    "lauren-docs": {
      "type": "http",
      "url": "https://lauren-py.dev/api/mcp"
    }
  }
}

Add --scope user to enable it in every project, or --scope project to write the .mcp.json shown above.
Run /mcp inside Claude Code to check that lauren-docs is connected.

What your agent gets

Four read-only tools. Agents usually search first, then read the exact section they need.

search_docs

Search Lauren docs

Ranked full-text search over every documentation page and the llms-full.txt API reference. Returns paths, URLs, the best-matching section and a snippet.

search_docs { "query": "websocket guards" }

get_doc

Read a Lauren docs page

The full Markdown of one page, or just one section via heading. Accepts paths like guides/websockets, /docs/… URLs and full site URLs.

get_doc { "path": "guides/websockets", "heading": "Authorisation" }

list_docs

List Lauren docs pages

The table of contents: every page grouped by section, with its title, path and URL.

list_docs { "section": "guides" }

get_reference

Get the Lauren API reference

llms-full.txt (the complete, condensed public-API reference written for LLMs) or the short llms.txt overview.

get_reference { "variant": "full" }

Every page is also an MCP resource — lauren-docs://docs/{path} — next to lauren-docs://llms.txt and lauren-docs://llms-full.txt, so clients with a resource picker can attach docs directly.

Try it

Ask for a change in plain language and mention the docs — the agent does the lookups.

Use the lauren-docs MCP server to look up how guards work on WebSocket gateways, then add a @ws_controller chat gateway that reuses our HTTP AuthGuard, with a WsTestClient test.

Search the Lauren docs for sessions and add login / logout endpoints that call regenerate_id() on login.

Read the Lauren background-tasks guide and move our welcome e-mail into BackgroundTasks so the signup endpoint returns immediately.

Tip: add a line such as “For anything about the Lauren framework, consult the lauren-docs MCP server before writing code” to your CLAUDE.md, AGENTS.md or editor rules so the agent reaches for it unprompted.

Always current

The index is rebuilt on every deploy from the same Markdown that renders these docs — no stale training data.

Read-only and anonymous

No API key, no account, no sessions. The tools can only read the public documentation.

Standard transport

Streamable HTTP with JSON responses, MCP protocol versions 2025-03-26 through 2025-11-25.

Check the endpoint

Any JSON-RPC request works — this one lists the tools. For an interactive view, run npx @modelcontextprotocol/inspector and connect to the URL with the Streamable HTTP transport.

Terminal

curl -s https://lauren-py.dev/api/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Prefer plain files?

The same reference ships as static files following the llms.txt convention — handy for tools without MCP support:

/llms.txt — short overview with links
/llms-full.txt — the complete condensed API reference

Reading it yourself?

The agent reads exactly what you see in the documentation.

Browse the docs Agent skills for Lauren