# Docs MCP (https://crawler-docs.lumen-labs.ai/docs/v2/docs-mcp)

> For the complete Lumen Crawler documentation index, see [llms.txt](https://crawler-docs.lumen-labs.ai/docs/llms.txt). A markdown version of any docs page is available by appending `.md` to its URL (e.g. https://crawler-docs.lumen-labs.ai/docs/v2/docs-mcp.md).

Connect coding agents to Lumen Crawler documentation over Streamable HTTP.

This is the **documentation** MCP — search and fetch public docs as markdown. It does not create crawls. Crawl HTTP is documented in the [quickstart](https://crawler-docs.lumen-labs.ai/docs/v2/quickstart).

**Endpoint:** `https://crawler-docs.lumen-labs.ai/docs/mcp`  
**Discovery:** [/docs/.well-known/mcp.json](/.well-known/mcp.json)  
**Machine index:** [/docs/llms.txt](/llms.txt) · append `.md` to any docs URL for markdown

## Tools

| Tool | Purpose |
| --- | --- |
| `search_docs` | Full-text search |
| `list_sections` | List published pages |
| `get_page` | Fetch a page as markdown |
| `get_code_example` | Extract the first (or language-matched) fenced code block |

## Connect

Copy the JSON from [Copy-paste prompts](https://crawler-docs.lumen-labs.ai/docs/v2/prompts). No crawl API key is required.

## Agent workflow

1. Start from [/docs/llms.txt](/llms.txt) or `list_sections`.
2. `get_page` with a path like `/v2/quickstart`.
3. For API calls: POST **202** + `id`; poll GET until a terminal status; **404** means unknown crawl, not pending.
4. Optional skill: [/docs/skill.md](/skill.md)

Local docs server: `http://localhost:3001/docs/mcp` after `npm run dev`.