@mcpframework/docs

Spin up an MCP documentation server from your Fumadocs site (or any site with llms.txt) in minutes. AI agents in Claude Code, Cursor, or any MCP client get tools to search, browse, and retrieve your documentation — enabling them to write correct integration code on the first try.

Built on top of mcp-framework.

Quick Start

import { DocsServer, FumadocsRemoteSource } from "@mcpframework/docs";

const source = new FumadocsRemoteSource({
  baseUrl: "https://docs.myapi.com",
});

const server = new DocsServer({
  source,
  name: "my-api-docs",
  version: "1.0.0",
});

server.start();

Or scaffold a project instantly:

npx create-docs-mcp my-api-docs

Sources

FumadocsRemoteSource

Purpose-built for Fumadocs sites. Uses the native Orama search API for high-quality results, with automatic fallback to local text search.

import { FumadocsRemoteSource } from "@mcpframework/docs";

const source = new FumadocsRemoteSource({
  baseUrl: "https://docs.myapi.com",     // Required
  searchEndpoint: "/api/search",          // Default: "/api/search"
  llmsTxtPath: "/llms.txt",              // Default: "/llms.txt"
  llmsFullTxtPath: "/llms-full.txt",     // Default: "/llms-full.txt"
  refreshInterval: 300_000,              // Cache TTL in ms (default: 5 min)
  headers: {                              // Optional custom headers
    Authorization: "Bearer ...",
  },
});

Option	Type	Default	Description
baseUrl	string	required	Base URL of your Fumadocs site
searchEndpoint	string	"/api/search"	Fumadocs Orama search endpoint
llmsTxtPath	string	"/llms.txt"	Path to llms.txt index
llmsFullTxtPath	string	"/llms-full.txt"	Path to full content
refreshInterval	number	300000	Cache TTL in milliseconds
headers	Record<string, string>	undefined	Custom HTTP headers

LlmsTxtSource

Works with any documentation site that publishes llms.txt and llms-full.txt files (Fumadocs, Docusaurus with plugin, etc.). Search is performed locally via text matching.

import { LlmsTxtSource } from "@mcpframework/docs";

const source = new LlmsTxtSource({
  baseUrl: "https://docs.myapi.com",
  mdxPathPrefix: "/docs/",              // Default: "/"
  refreshInterval: 300_000,
});

Option	Type	Default	Description
baseUrl	string	required	Base URL of your docs site
llmsTxtPath	string	"/llms.txt"	Path to llms.txt index
llmsFullTxtPath	string	"/llms-full.txt"	Path to full content
mdxPathPrefix	string	"/"	Prefix for individual page .mdx fetching
refreshInterval	number	300000	Cache TTL in milliseconds
headers	Record<string, string>	undefined	Custom HTTP headers
cache	Cache	MemoryCache	Custom cache implementation

Tools

The server exposes three MCP tools:

search_docs

Search documentation by keyword or phrase. Returns ranked results with excerpts.

Parameter	Type	Required	Description
query	string	yes	Search keywords or phrase
section	string	no	Filter to a specific section
limit	number	no	Max results (default 10, max 25)

get_page

Retrieve the full markdown content of a documentation page.

Parameter	Type	Required	Description
slug	string	yes	Page slug or URL path

list_sections

Browse the documentation tree structure to discover available content.

Parameter	Type	Required	Description
section	string	no	Filter to a section's children

Fumadocs Setup

Your Fumadocs site needs to serve these endpoints:

1. /llms.txt (required) — Generated by fumadocs-core's source.llms().index() utility
2. /llms-full.txt (required for search) — Generated by source.llms().full()
3. /api/search (optional) — Fumadocs' built-in Orama search API for better results

See the Fumadocs LLMs.txt documentation for setup instructions.

MCP Client Configuration

Claude Code

claude mcp add my-api-docs -- node /path/to/server/dist/index.js

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "my-api-docs": {
      "command": "node",
      "args": ["/path/to/server/dist/index.js"],
      "env": {
        "DOCS_BASE_URL": "https://docs.myapi.com"
      }
    }
  }
}

Cursor

Add to your MCP settings:

{
  "my-api-docs": {
    "command": "node",
    "args": ["/path/to/server/dist/index.js"],
    "env": {
      "DOCS_BASE_URL": "https://docs.myapi.com"
    }
  }
}

Custom Source Adapters

Implement the DocSource interface to create adapters for any documentation backend:

import { DocSource, DocPage, DocSearchResult, DocSection } from "@mcpframework/docs";

class MyCustomSource implements DocSource {
  name = "my-source";

  async search(query: string, options?: { section?: string; limit?: number }) {
    // Your search implementation
    return [];
  }

  async getPage(slug: string) {
    // Fetch and return a page, or null
    return null;
  }

  async listSections() {
    // Return your documentation tree
    return [];
  }

  async getIndex() {
    return ""; // llms.txt content
  }

  async getFullContent() {
    return ""; // llms-full.txt content
  }

  async healthCheck() {
    return { ok: true };
  }
}

Caching

The built-in MemoryCache provides LRU eviction with TTL expiry. You can pass a custom cache:

import { MemoryCache, LlmsTxtSource } from "@mcpframework/docs";

const cache = new MemoryCache({
  maxEntries: 200,        // Default: 100
  ttlMs: 600_000,         // Default: 300_000 (5 min)
});

const source = new LlmsTxtSource({
  baseUrl: "https://docs.myapi.com",
  cache,
});

Or implement the Cache interface for Redis, SQLite, etc.

API

// Main exports
import {
  DocsServer,
  LlmsTxtSource,
  FumadocsRemoteSource,
  MemoryCache,
  SearchDocsTool,
  GetPageTool,
  ListSectionsTool,
} from "@mcpframework/docs";

// Subpath imports
import { LlmsTxtSource, FumadocsRemoteSource } from "@mcpframework/docs/sources";
import { SearchDocsTool, GetPageTool, ListSectionsTool } from "@mcpframework/docs/tools";
import { MemoryCache } from "@mcpframework/docs/cache";

License

MIT