docs-indexer-mcp

k2sebeom/docs-indexer-mcp

3.2

If you are the rightful owner of docs-indexer-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

Docs Indexer MCP is a versatile tool designed to index and read documentation, functioning both as a standalone CLI application and as an MCP server for AI assistants.

Tools

  1. list_documentations

    List all available documentations.

  2. list_pages

    List available pages in a documentation.

  3. read_page

    Read a specific page from documentation.

Docs Indexer MCP

A documentation indexing and reading tool that works both as a standalone CLI application and as an MCP (Model Context Protocol) server for AI assistants.

Overview

Docs Indexer MCP allows you to:

  1. Crawl and index web-based documentation
  2. List available documentation sets
  3. Browse pages within documentation
  4. Read documentation content in markdown format
  5. Integrate with AI assistants via MCP

Installation

# Clone the repository
git clone https://github.com/yourusername/docs-indexer-mcp.git
cd docs-indexer-mcp

uv run cli

Usage

CLI Mode

The tool can be used as a standalone command-line application:

# Start the CLI
uv run cli

Available commands:

  • help - Show help information
  • exit - Exit the program
  • crawl <title> <url> <prefix> - Crawl and index a documentation
  • list - List all available documentations
  • pages <doc_name> - List all pages in a documentation
  • read <doc_name> <page_number> - Read a specific page from documentation

Example workflow:

# Crawl and index Python documentation
>>> crawl python https://docs.python.org/3/ https://docs.python.org/3/

# List available documentations
>>> list

# List pages in Python documentation
>>> pages python

# Read a specific page (page number 5)
>>> read python 5

MCP Server Mode

The tool can also run as an MCP server to provide documentation access to AI assistants:

# Start the MCP server
uv run docs-indexer-mcp

When running as an MCP server, the following tools are available to AI assistants:

  • list_documentations - List all available documentations
  • list_pages - List available pages in a documentation
  • read_page - Read a specific page from documentation

Data Storage

All indexed documentation is stored in ~/.docs_indexer/docs/ with the following structure:

~/.docs_indexer/
  └── docs/
      └── <doc_name>/
          └── meta.json

Each meta.json file contains:

  • Documentation name
  • Base URL
  • URL prefix for crawling
  • List of pages with titles and URLs
  • Last sync timestamp

Features

  • Web Crawling: Automatically crawls documentation websites to index pages
  • HTML to Markdown: Converts HTML documentation to readable markdown format
  • MCP Integration: Works with AI assistants that support the Model Context Protocol
  • Local Storage: Stores indexed documentation locally for offline access

Requirements

  • Python 3.13+
  • Dependencies:
    • beautifulsoup4
    • html2text
    • mcp[cli]
    • requests