meilisearch-documentation-mcp

JuniorJoanis/meilisearch-documentation-mcp

3.2

If you are the rightful owner of meilisearch-documentation-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 dayong@mcphub.com.

The Meilisearch Documentation MCP Server is a production-ready server that provides structured access to Meilisearch documentation for LLMs.

Tools
3
Resources
0
Prompts
0

Meilisearch Documentation MCP Server

A production-ready Model Context Protocol (MCP) server that exposes the Meilisearch documentation as structured tools, resources, and prompts for LLMs.

Features

  • Full-text search across all Meilisearch documentation
  • Resource access to individual documentation pages
  • Prompt templates for common documentation queries
  • Local indexing for fast, deterministic searches
  • Automatic repository management with caching

Installation

pip install -e .

Or install dependencies directly:

pip install mcp mistune python-frontmatter whoosh GitPython pydantic

Usage

Running the MCP Server

python -m meilisearch_docs_mcp

The server will:

  1. Clone the Meilisearch documentation repository (if not already cached)
  2. Parse all MDX files
  3. Build a local search index
  4. Start the MCP server on stdio

Configuration

The server can be configured via environment variables:

  • MEILISEARCH_DOCS_REPO_URL: Repository URL (default: https://github.com/meilisearch/documentation.git)
  • MEILISEARCH_DOCS_CACHE_DIR: Cache directory for cloned repo (default: .cache/docs)
  • MEILISEARCH_DOCS_INDEX_DIR: Directory for search index (default: .cache/index)

MCP Integration

For Cursor

Add to your Cursor MCP settings (usually in .cursor/mcp.json or Cursor settings):

{
  "mcpServers": {
    "meilisearch-docs": {
      "command": "python",
      "args": ["-m", "meilisearch_docs_mcp"],
      "env": {}
    }
  }
}

For Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "meilisearch-docs": {
      "command": "python",
      "args": ["-m", "meilisearch_docs_mcp"]
    }
  }
}

Note: Make sure to use the full path to your Python interpreter if it's not in your PATH, e.g.:

{
  "mcpServers": {
    "meilisearch-docs": {
      "command": "/usr/local/bin/python3",
      "args": ["-m", "meilisearch_docs_mcp"]
    }
  }
}

For ChatGPT

Configure via the MCP settings in your ChatGPT interface. The server communicates via stdio using the MCP protocol.

Available Tools

  • search_documentation(query: str, limit: int = 10) - Search the documentation with a query string
  • get_documentation_section(section: str) - Get all pages from a specific section (guides/reference/learn)
  • list_documentation_sections() - List all available documentation sections

Available Resources

Resources are accessible via URI pattern: meilisearch-docs://{path}

Examples:

  • meilisearch-docs://guides/getting_started/quick_start_guide.mdx
  • meilisearch-docs://reference/api/search.mdx

Available Prompts

  • get_started - Get started with Meilisearch
  • api_reference - Query API reference documentation
  • troubleshooting - Troubleshoot common issues
  • integration_guide - Find integration-specific guides

Development

# Clone the repository
git clone <repository-url>
cd meilisearch-documentation-mcp

# Install in development mode
pip install -e .

# Or install dependencies directly
pip install mcp mistune python-frontmatter whoosh GitPython pydantic

# Run the server
python -m meilisearch_docs_mcp

How It Works

  1. Repository Management: On first run, the server clones the Meilisearch documentation repository to a local cache directory (.cache/docs/meilisearch-documentation).

  2. Parsing: All MDX files are parsed to extract:

    • Frontmatter metadata (title, description, etc.)
    • Markdown content
    • Section classification (guides/reference/learn)
  3. Indexing: A local Whoosh full-text search index is built from the parsed documents, stored in .cache/index/meilisearch_docs.

  4. MCP Server: The server exposes:

    • Tools: Search and query the documentation
    • Resources: Access individual documentation pages by URI
    • Prompts: Pre-configured prompts for common documentation queries

Troubleshooting

Index Not Found Error

If you see an error about the index not existing, the server will automatically rebuild it. This may take a few minutes on first run.

Repository Clone Fails

  • Check your internet connection
  • Verify the repository URL is accessible
  • Ensure you have write permissions in the cache directory
  • Try deleting .cache/docs/meilisearch-documentation and restarting

Search Returns No Results

  • Verify the index was built successfully (check logs)
  • Try rebuilding the index by deleting .cache/index/meilisearch_docs
  • Check that MDX files were parsed correctly

License

MIT