📚 Expo React Native Markdown Documentation Server (MCP)

A fast, extensible, and intelligent documentation server for Markdown files using React Native Expo Documentation, powered by the MCP framework.

🚀 Features

• Automatic Markdown Discovery: Scans the docs/ directory for all Markdown files and registers them as searchable resources.
• Metadata Extraction: Reads YAML frontmatter for tags and versioning, and tracks last modified dates.
• Powerful Search Tools:
  • Keyword Search: Find documentation by exact keyword matches, with ranked results and context snippets.
  • Fuzzy Search: Typo-tolerant search using rapidfuzz for partial matches.
  • Advanced Search: Filter by keyword, tag, and/or last modified date.
• Resource API: Exposes documentation as resources with rich metadata for integration with other tools.
• Logging: All queries are logged for analytics and debugging.
• Simple API: Exposes all search and retrieval tools via MCP's tool interface.

🏗️ Project Structure

.
├── server.py         # Main server code (this file)
├── docs/            # Place your Markdown documentation here
│   └── *.md
├── sitemap_scrape.py  # Scrapes Expo and React Native Sitemaps for Documentation URLS and saves to sitemaps_urls.json
├── scrape_markdown.py  # Scrapes sitemaps_urls.json URLs and saves the data as *.md files in /docs
└── README.md

🛠️ Usage

1. Add Your Documentation

Place your Markdown files in the docs/ directory or use sitemap_scrape.py and scrape_markdown.py to generate *.md files from a sitemap url.

Optionally, add YAML frontmatter for tags and versioning:

---
tags: [api, quickstart]
version: 1.2.0
---

# My API Documentation

Welcome to the docs!

2. Start the Server

Note: use source .venv/bin/activate to activate the python enviornment

 uv run mcp dev server.py 

The server runs using MCP's stdio transport by default.

🔍 API Overview

All tools are exposed via the MCP tool interface:

search_docs(keyword: str)

• Description: Search for a keyword in all docs. Returns ranked filenames with context snippets.
• Returns:
  • filename
  • matches (count)
  • snippet
  • last_modified
  • tags
  • version

fuzzy_search_docs(keyword: str, threshold: int = 70)

• Description: Typo-tolerant search using rapidfuzz. Returns ranked results with context.
• Returns:
  • filename
  • score
  • snippet
  • last_modified
  • tags
  • version

advanced_search(keyword: str = "", tag: str = "", after: str = "")

• Description: Search by keyword, tag, or last modified date (ISO format).
• Returns:
  • filename
  • matches
  • snippet
  • last_modified
  • tags
  • version

get_doc_content(filename: str)

• Description: Retrieve the full content of a documentation file by name.

📝 Example Query

from mcp.client import MCPClient

client = MCPClient("http://localhost:YOUR_PORT")
results = client.search_docs("installation")
print(results)

🍾 Adding to Cursor

Use the mcp.json file structure for using the MCP server in Cursor

🧩 Extending

• Add new Markdown files to docs/ — they are auto-discovered.
• Add new search tools or resource types by extending server.py.

🛡️ Logging

All queries are logged to mcp_server_queries.log for traceability and analytics.

🧑‍💻 Requirements

• Python 3.8+
• mcp
• PyYAML
• rapidfuzz

Install dependencies:

pip install mcp pyyaml rapidfuzz

🤝 Contributing

Pull requests and issues are welcome!
Feel free to suggest features or improvements.

📄 License

MIT License

✨ Credits

Built with MCP and ❤️ by yours truly.