Gmail MCP Server 15s

🚀 Gmail MCP Server

An integrated Gmail system with FastMCP (Model Control Protocol) that enables AI assistants to interact with Gmail through a series of tools and prompts.

🚀 Quick Start

This project provides a server-client architecture that allows AI assistants to perform the following Gmail operations:

• Send emails
• View unread emails
• Browse email content
• Move emails to the trash
• Open emails in the browser
• Mark emails as read

It uses OAuth 2.0 for authentication and integrates with the Gmail API to achieve these functions.

✨ Features

Gmail Server

Responsible for:

• FastMCP integration
• Authentication
• Email operations (sending, receiving, etc.)
• Error handling and logging

Smart Gmail Client

Utilizes an LLM (e.g., Gemini 1.5 Flash) to understand natural language requests, extract parameters, and execute corresponding email operations.

📦 Installation

Prerequisites

The following environment is required:

• Python 3.6 or higher
• Install the necessary libraries: google-auth, google-api-python-client, fastmcp

Prepare the following files:

• credentials.json: OAuth credential file generated by the Google Developer Console.
• token.json: Stores the OAuth authentication token.
• .env: Contains API keys and relevant configurations.

💻 Usage Examples

Start the Server

python3 gmail_server.py

Use the Client

from gmail_client import GmailClient
client = GmailClient()
result = client.send_email(to="recipient@example.com", subject="Test", message="Hello World!")
print(result)

🔧 Technical Details

Email Content Processing

Supports MIME message parsing, multi-part emails, and attachment handling to ensure correct decoding of email headers and bodies.

LLM Integration

Uses Gemini 1.5 Flash to convert natural language requests into structured tool calls and handles non-standard responses through a fallback mechanism.

📚 Documentation

Development and Extension

To add new Gmail operations:

1. Implement new tools in gmail_server.py.
2. The system will automatically discover and expose new features.

Customize LLM Prompts

You can optimize parameter extraction and specific request handling by modifying the prompts in test_gmail_server.py.

Security Precautions

• The project does not include credential files. Do not commit the following files to the version control system:
  • credentials.json
  • token.json
  • .env

These files have been added to .gitignore to prevent accidental commits.

📄 References

Medium Article