Docs Search Engine

A custom Model Context Protocol (MCP) server that acts as a documentation search engine.

This project attempts to build a simple, personal clone of Context7, unlocking the capability to access up-to-date documentation from GitHub repositories and web pages directly within your AI assistant's context.

🛠️ Tech Stack

• Python: Core programming language.
• FastMCP: Framework for building MCP servers easily.
• minsearch: Lightweight, in-memory full-text search engine.
• uv: Fast Python package and environment manager.
• Jina Reader: For turning web pages into LLM-friendly markdown.
• requests: For handling HTTP requests and downloading zip files.
• pytest: For comprehensive testing.

📂 Project Structure

docs-search-engine/
├── main.py             # Entry point: Defines MCP tools and server configuration
├── search.py           # Core logic: Zip download, extraction, indexing, and search
├── scrape_web.py       # Web scraping functionality (using Jina Reader)
├── test_search.py      # Tests for search functionality
├── test_scrape_web.py  # Tests for web scraping
└── pyproject.toml      # Project dependencies and configuration

🚀 Workflow

1. Ingestion: The server downloads documentation source code (e.g., as a .zip from GitHub).
2. Indexing: Markdown content (.md and .mdx) is extracted and indexed in-memory using minsearch.
3. Caching: Indexes are cached by URL to ensure fast subsequent searches without re-downloading.
4. Retrieval: Users query the system via MCP tools (search_docs, scrape_web), and relevant context is returned to the LLM.

⚙️ MCP Configuration

Add the following configuration to your MCP client settings (e.g., mcp_config.json in Google Antigravity):

{
  "mcpServers": {
    "docs-search-engine": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "C:/Users/username/path/to/docs-search-engine",
        "main.py"
      ]
    }
  }
}

Note: Replace C:/Users/username/path/to/docs-search-engine with the actual absolute path to your project directory.

💡 Example Usage

Once the MCP server is connected to your AI assistant (e.g., VSCode, Claude, Cursor, Antigravity), you can use natural language to interact with it.

1. Search Documentation

"Search for 'context' in the FastMCP docs."

"Find information about 'indexing' in the minsearch docs (https://github.com/alexeygrigorev/minsearch)."

2. Scrape Web Pages

"Scrape the content of https://example.com/blog/article and summarize it."

3. Count Word Occurrences

"Count how many times the word 'LLM' appears on https://example.com/ai-trends."

💻 Setup & Execution

Prerequisites

• Python 3.13+
• uv installed (recommended)

Installation

1. Clone the repository and navigate to the directory:

   cd docs-search-engine
   

2. Install dependencies:

   uv sync
   

Running Locally

To run the server manually for debugging:

uv run main.py

Testing

Run the comprehensive test suite to ensure everything is working correctly:

# Run all tests
uv run pytest -v

# Run specific test files
uv run pytest test_search.py -v
uv run pytest test_scrape_web.py -v

# Run only integration tests
uv run pytest -m integration -v