Meilisearch MCP Server

A Model Context Protocol (MCP) server for interacting with Meilisearch through LLM interfaces like Claude.

Features

• Index and document management
• Settings configuration and management
• Task monitoring and API key management
• Built-in logging and monitoring tools
• Dynamic connection configuration to switch between Meilisearch instances
• Smart search across single or multiple indices
• This is a Python implementation, there is Typescript integration if you need to work with a Meilisearch MCP server within the browser

Installation

# Clone repository
git clone <repository_url>
cd meilisearch-mcp

# Create virtual environment and install
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

# Install development dependencies (for testing and development)
uv pip install -r requirements-dev.txt

Requirements

• Python ≥ 3.9
• Running Meilisearch instance
• Node.js (for testing with MCP Inspector)

Development Setup

Prerequisites

1. Start Meilisearch server:

   # Using Docker (recommended for development)
   docker run -d -p 7700:7700 getmeili/meilisearch:v1.6
   
   # Or using brew (macOS)
   brew install meilisearch
   meilisearch
   
   # Or download from https://github.com/meilisearch/meilisearch/releases

2. Install development tools:

   # Install uv for Python package management
   pip install uv
   
   # Install Node.js for MCP Inspector testing
   # Visit https://nodejs.org/ or use your package manager

Running Tests

This project includes comprehensive integration tests that verify MCP tool functionality:

# Run all tests
python -m pytest tests/ -v

# Run specific test file
python -m pytest tests/test_mcp_client.py -v

# Run tests with coverage report
python -m pytest --cov=src tests/

# Run tests in watch mode (requires pytest-watch)
pytest-watch tests/

Important: Tests require a running Meilisearch instance on http://localhost:7700. The tests will:

• Create temporary test indices with unique names
• Test all MCP tools end-to-end
• Clean up test data automatically
• Verify error handling and edge cases

Code Quality

# Format code with Black
black src/ tests/

# Run type checking (if mypy is configured)
mypy src/

# Lint code (if flake8 is configured)
flake8 src/ tests/

Usage

Environment Variables

MEILI_HTTP_ADDR=http://localhost:7700  # Default Meilisearch URL
MEILI_MASTER_KEY=your_master_key       # Optional: Default Meilisearch API key

Dynamic Connection Configuration

The server provides tools to view and update connection settings at runtime:

• get-connection-settings: View current connection URL and API key status
• update-connection-settings: Update URL and/or API key to connect to a different Meilisearch instance

Example usage through MCP:

// Get current settings
{
  "name": "get-connection-settings"
}

// Update connection settings
{
  "name": "update-connection-settings",
  "arguments": {
    "url": "http://new-host:7700",
    "api_key": "new-api-key"
  }
}

Search Functionality

The server provides a flexible search tool that can search across one or all indices:

• search: Search through Meilisearch indices with optional parameters

Example usage through MCP:

// Search in a specific index
{
  "name": "search",
  "arguments": {
    "query": "search term",
    "indexUid": "movies",
    "limit": 10
  }
}

// Search across all indices
{
  "name": "search",
  "arguments": {
    "query": "search term",
    "limit": 5,
    "sort": ["releaseDate:desc"]
  }
}

Available search parameters:

• query: The search query (required)
• indexUid: Specific index to search in (optional)
• limit: Maximum number of results per index (optional, default: 20)
• offset: Number of results to skip (optional, default: 0)
• filter: Filter expression (optional)
• sort: Sorting rules (optional)

Running the Server

python -m src.meilisearch_mcp

Usage with Claude Desktop

To use this with Claude Desktop, add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "meilisearch": {
      "command": "uvx",
      "args": ["-n", "meilisearch-mcp"]
    }
  }
}

Testing with MCP Inspector

npx @modelcontextprotocol/inspector python -m src.meilisearch_mcp

Available Tools

Connection Management

• get-connection-settings: View current Meilisearch connection URL and API key status
• update-connection-settings: Update URL and/or API key to connect to a different instance

Index Management

• create-index: Create a new index with optional primary key
• list-indexes: List all available indexes
• delete-index: Delete an existing index and all its documents
• get-index-metrics: Get detailed metrics for a specific index

Document Operations

• get-documents: Retrieve documents from an index with pagination
• add-documents: Add or update documents in an index

Search

• search: Flexible search across single or multiple indices with filtering and sorting options

Settings Management

• get-settings: View current settings for an index
• update-settings: Update index settings (ranking, faceting, etc.)

API Key Management

• get-keys: List all API keys
• create-key: Create new API key with specific permissions
• delete-key: Delete an existing API key

Task Management

• get-task: Get information about a specific task
• get-tasks: List tasks with optional filters:
  • limit: Maximum number of tasks to return
  • from: Number of tasks to skip
  • reverse: Sort order of tasks
  • batchUids: Filter by batch UIDs
  • uids: Filter by task UIDs
  • canceledBy: Filter by cancellation source
  • types: Filter by task types
  • statuses: Filter by task statuses
  • indexUids: Filter by index UIDs
  • afterEnqueuedAt/beforeEnqueuedAt: Filter by enqueue time
  • afterStartedAt/beforeStartedAt: Filter by start time
  • afterFinishedAt/beforeFinishedAt: Filter by finish time
• cancel-tasks: Cancel pending or enqueued tasks
• delete-tasks: Delete completed tasks

System Monitoring

• health-check: Basic health check
• get-health-status: Comprehensive health status
• get-version: Get Meilisearch version information
• get-stats: Get database statistics
• get-system-info: Get system-level information

Contributing

We welcome contributions! Please follow these guidelines:

1. Fork and clone the repository
2. Set up development environment following the Development Setup section above
3. Create a feature branch from main
4. Write tests first if adding new functionality (Test-Driven Development)
5. Run tests locally to ensure all tests pass before committing
6. Format code with Black and ensure code quality
7. Commit changes with descriptive commit messages
8. Push to your fork and create a pull request

Development Workflow

# Create feature branch
git checkout -b feature/your-feature-name

# Make your changes, write tests first
# Edit files...

# Run tests to ensure everything works
python -m pytest tests/ -v

# Format code
black src/ tests/

# Commit and push
git add .
git commit -m "Add feature description"
git push origin feature/your-feature-name

Testing Guidelines

• All new features should include tests
• Tests should pass before submitting PRs
• Use descriptive test names and clear assertions
• Test both success and error cases
• Ensure Meilisearch is running before running tests

Release Process

This project uses automated versioning and publishing to PyPI. The release process is designed to be simple and automated.

How Releases Work

1. Automated Publishing: When the version number in pyproject.toml changes on the main branch, a GitHub Action automatically:

   • Builds the Python package
   • Publishes it to PyPI using trusted publishing
   • Creates a new release on GitHub

2. Version Detection: The workflow compares the current version in pyproject.toml with the previous commit to detect changes

3. PyPI Publishing: Uses PyPA's official publish action with trusted publishing (no manual API keys needed)

Creating a New Release

To create a new release, follow these steps:

1. Determine Version Number

Follow Semantic Versioning (MAJOR.MINOR.PATCH):

• PATCH (e.g., 0.4.0 → 0.4.1): Bug fixes, documentation updates, minor improvements
• MINOR (e.g., 0.4.0 → 0.5.0): New features, new MCP tools, significant enhancements
• MAJOR (e.g., 0.5.0 → 1.0.0): Breaking changes, major API changes

2. Update Version and Create PR

# 1. Create a branch from latest main
git checkout main
git pull origin main
git checkout -b release/v0.5.0

# 2. Update version in pyproject.toml
# Edit the version = "0.4.0" line to your new version

# 3. Commit and push
git add pyproject.toml
git commit -m "Bump version to 0.5.0"
git push origin release/v0.5.0

# 4. Create PR and get it reviewed/merged
gh pr create --title "Release v0.5.0" --body "Bump version for release"

3. Merge to Main

Once the PR is approved and merged to main, the GitHub Action will automatically:

1. Detect the version change
2. Build the package
3. Publish to PyPI at https://pypi.org/p/meilisearch-mcp
4. Make the new version available via pip install meilisearch-mcp

4. Verify Release

After merging, verify the release:

# Check GitHub Action status
gh run list --workflow=publish.yml

# Verify on PyPI (may take a few minutes)
pip index versions meilisearch-mcp

# Test installation of new version
pip install --upgrade meilisearch-mcp

Release Workflow File

The automated release is handled by .github/workflows/publish.yml, which:

• Triggers on pushes to main branch
• Checks if pyproject.toml version changed
• Uses Python 3.10 and official build tools
• Publishes using trusted publishing (no API keys required)
• Provides verbose output for debugging

Troubleshooting Releases

Release didn't trigger: Check that the version in pyproject.toml actually changed between commits

Build failed: Check the GitHub Actions logs for Python package build errors

PyPI publish failed: Verify the package name and that trusted publishing is configured properly

Version conflicts: Ensure the new version number hasn't been used before on PyPI

Development vs Production Versions

• Development: Install from source using pip install -e .
• Production: Install from PyPI using pip install meilisearch-mcp
• Specific version: Install using pip install meilisearch-mcp==0.5.0

License

MIT