MCP Google Calendar Server with Persistent Auth

This project implements a Model Context Protocol (MCP) server for Google Calendar, featuring persistent OAuth 2.0 authentication using Redis. It allows AI assistants (like Claude or Custom Clients) to interact with your Google Calendar.

Features

• Google Calendar Integration: List upcoming events and create new events.

• Persistent Authentication: Uses Redis to store OAuth tokens securely, so you don't need to re-authenticate on every restart.

• FastMCP Framework: Built on top of the modern FastMCP framework.

• Client Implementation: Includes a sample client (client.py) using OpenAI's GPT-4o to demonstrate interaction.

Related MCP server: Google Calendar MCP Server

Supported Tools

1. list_events: Lists upcoming events from the user's primary calendar. Supports filtering by result count and minimum time.

2. create_event: Creates a new event with a summary, start time, end time, and description.

Prerequisites

Before running this project, ensure you have the following:

1. Python 3.12+: This project requires a recent version of Python.

2. Redis Server: A running Redis instance is required for token storage.

3. Google Cloud Project:

   • Enable the Google Calendar API.

   • Create OAuth 2.0 Client IDs (Desktop app).

   • Scopes: Ensure the client has access to https://www.googleapis.com/auth/calendar.events.

   • Redirect URI: If configured for web, ensure http://localhost:8000/auth/callback is allowed (FastMCP default). For Desktop/Installed App, this is handled automatically.

   • Download the JSON credentials or copy the Client ID and Client Secret.

4. OpenAI API Key: Required if you want to run the provided client.py test script.

Installation

This project manages dependencies using uv (recommended) or you can use pip.

Using uv (Recommended)

1. Sync the project environment:

   uv sync

Using pip

1. Create a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

2. Install dependencies:

   pip install -e .

Configuration

1. Copy the example environment file:

   cp .env.example .env
   # On Windows: copy .env.example .env

2. Edit .env and fill in your credentials:

Please refer .env.example

Docker Support

You can also run the server using Docker.

1. Build the image:

   docker build -t mcp-google-calendar .

2. Run the container (ensure you environment variables are passed):

   # Assuming you have a .env file
   docker run --env-file .env -p 8000:8000 mcp-google-calendar

   Note: If your Redis is running on the host, use host.docker.internal (Windows/Mac) or --network="host" (Linux).

Running the Server

1. Start Redis: Ensure your Redis server is running locally (default port 6379).

2. Run the MCP Server: Using uv:

   uv run server.py

   Using standard python:

   python server.py

   The server will start at http://localhost:8000.

You can also use profiles to maintain separate authentication sessions:

uv run client.py --profile work
uv run client.py --profile personal
uv run client.py --profile user1
uv run client.py --profile user2

Connecting a Client

Option 1: Using the provided Test Client

This project includes a CLI client powered by OpenAI to test the integration.

1. Ensure the server is running.

2. In a new terminal, run:

   uv run client.py
   # or: python client.py

3. On first run, it will open your request OAuth authentication. Follow the steps in your browser to log in with your Google account.

4. Once authenticated, you can chat with the assistant:

   "What is on my calendar for today?" "Schedule a meeting with the team tomorrow at 2 PM."

Option 2: Using Claude Desktop

You can configure Claude Desktop to use this MCP server.

1. Locate your Claude Desktop config file (typically %APPDATA%\Claude\claude_desktop_config.json on Windows or ~/Library/Application Support/Claude/claude_desktop_config.json on macOS).

2. Add the server configuration:

   {
     "mcpServers": {
       "google-calendar": {
         "command": "uv",
         "args": ["run", "--directory", "C:/path/to/repo/mcp-auth-google-calendar", "server.py"],
         "env": {
            "GOOGLE_CLIENT_ID": "your_client_id",
            "GOOGLE_CLIENT_SECRET": "your_client_secret"
         }
       }
     }
   }

   Note: Replace paths and credentials with your actual values.

Testing

Currently, no automated unit tests are included. The best way to test is using the client.py script as described above.

To run future tests (if added), you would use pytest:

uv run pytest