Installation

Install and run the data.gouv.fr MCP server locally using uv

This guide walks you through installing and running the data.gouv.fr MCP server manually using the uv package manager.

Prerequisites

Before you begin, ensure you have:

• Python 3.13+ installed on your system
• uv package manager installed

Installation Steps

1. Clone the repository

   Clone the repository and navigate to the project directory:

   git clone [e-mail]:datagouv/datagouv-mcp.git
   cd datagouv-mcp
   
2. Install dependencies

   Use uv to install all required Python dependencies:

   uv sync
   

   This command reads the project configuration and installs all dependencies specified in pyproject.toml.

3. Configure environment variables

   Copy the example environment file to create your own configuration:

   cp .env.example .env
   

   The .env.example file contains all available configuration options with sensible defaults. You can edit the .env file to customize settings for your environment.

   Example .env configuration:

   MCP_HOST=127.0.0.1  # Bind to localhost for local development
   MCP_PORT=8000       # HTTP server port
   MCP_ENV=local       # Environment name for monitoring
   DATAGOUV_API_ENV=prod  # Use production data.gouv.fr API
   LOG_LEVEL=INFO      # Logging verbosity
   

   For local development, set MCP_HOST=127.0.0.1 to follow MCP security best practices. This ensures the server only accepts connections from your local machine.

   See the Configuration page for detailed information about all available environment variables.

4. Load environment variables

   Load the environment variables into your shell session:

   set -a && source .env && set +a
   

   This command exports all variables from the .env file to your current shell.

5. Start the server

   Run the MCP server:

   uv run main.py
   

   The server will start on the configured host and port (default: http://127.0.0.1:8000).

6. Verify the installation

   Test that the server is running by accessing the health endpoint:

   curl http://127.0.0.1:8000/health
   

   You should receive a JSON response like:

   {
     "status": "ok",
     "timestamp": "2026-03-04T12:34:56.789Z",
     "version": "1.0.0"
   }
   

Next Steps

Now that you have the server running:

• Connect a chatbot: Follow the connection instructions in the README to integrate with Claude, ChatGPT, or other MCP clients
• Configure advanced settings: See the Configuration page for all available options
• Deploy with Docker: Check out the Docker deployment guide for production deployments

Troubleshooting

Port already in use

If port 8000 is already in use, change the MCP_PORT in your .env file:

MCP_PORT=8007

Python version issues

Ensure you have Python 3.13 or higher:

python --version

Dependencies installation fails

Make sure uv is up to date:

pip install --upgrade uv