Gmail MCP Server

An MCP server integrating the Gmail API, supporting functions such as sending, deleting, reading, drafting, and replying to emails, which requires user authorization for operations.

Communication tools Artificial intelligence chatbots #Email management #Gmail integration #MCP service .Python

rating : 2.5 points

downloads : 7.6K

update time : 2025-04-29

Open Site

What is the Gmail MCP Server?

This is an intelligent service that connects to Gmail mailboxes, allowing you to manage emails through an AI assistant. It can help you automatically read new emails, draft replies, send emails, or organize your inbox, and all operations will first seek your confirmation.

How to use the Gmail MCP Server?

Simply complete a one - time Google API authorization setup, and then you can manage your Gmail mailbox with natural language instructions through supported clients (such as Claude Desktop).

Use cases

It is especially useful when you need an AI assistant to handle a large number of emails, such as quickly filtering important emails, automatically drafting common replies, and batch - organizing the inbox.

Main features

Send emails

Send new emails by specifying recipients, subjects, and content

Email management

Move emails to the trash or mark them as read

Read emails

Get a list of unread emails or read the content of specific emails

Browser integration

Open specific emails directly in the default browser

Advantages

The pre - operation confirmation mechanism ensures full user control

Seamless integration with Gmail, retaining all original functions

Supports mailbox operations through natural language instructions

Limitations

Requires a one - time complex API setup process

Only supports Gmail mailboxes

Some advanced Gmail functions may not be available

How to use

Create a Google Cloud project

Visit the Google Cloud console to create a new project

Enable the Gmail API

Search for and enable the Gmail API service in the API library

Configure the OAuth consent screen

Set the external user type and add test accounts

Create an OAuth client ID

Select the 'Desktop app' type and download the credential file

Configure the client

Modify the client configuration file to point to your credential and token file paths

Usage examples

Quickly process unread emails

Let the AI assistant list all unread emails, and you can choose to mark them as read or delete them

Automatically draft replies

Based on the received email content, let the AI assistant help draft a reply

Batch - organize the inbox

Process emails in batches according to conditions, such as deleting old notifications or marking important emails

Frequently Asked Questions

Is this service secure? Will it read all my emails?

Why are so many setup steps required?

Can I manage multiple Gmail accounts simultaneously?

How can I revoke the application's access permission?

Related resources

Google API Console

Create and manage Google Cloud projects

MCP Protocol Official Website

Learn detailed information about the Model Context Protocol

MCP Inspector Tool

A tool for testing and debugging MCP servers

uv Documentation

Official documentation for the server runtime environment uv

🚀 Gmail Server for Model Context Protocol (MCP)

This MCP server integrates with Gmail to enable functions such as sending, deleting, reading, drafting, and responding to emails.

⚠️ Important Note

This server allows MCP clients to read, delete, and send emails. However, the client will prompt the user before performing such activities.

🚀 Quick Start

✨ Features

This MCP server integrates with Gmail, enabling users to perform various email - related operations, including sending, deleting, reading, drafting, and responding to emails.

📦 Installation

Gmail API Setup

Create a new Google Cloud project
Enable the Gmail API
Configure the OAuth consent screen
- Select "External". However, we will not publish this application.
- Add your personal email address as a "Test user".
Add the OAuth scope https://www.googleapis.com/auth/gmail/modify
Create an OAuth client ID of the application type (e.g., "Desktop application")
Download your client's OAuth key JSON file
Rename the key file and save it to a secure location on your local computer. Note the location of this file.
- When starting the server, pass the absolute path as a parameter --creds-file-path.

Authentication

When the server starts, the authentication process will launch in your system browser. The credential token is then saved (and retrieved later) to the specified absolute file path, which is passed via the parameter --token-path.

For example, you can use the dot directory in your home directory:

Property	Details
`--creds-file-path`	`/[your-home-folder]/.google/client_creds.json`
`--token-path`	`/[your-home-folder]/.google/token.json`

Usage Instructions

To start the server, run the command:

python3 src/main.py --config config.yaml

Verification

To verify that the server is running properly, you can perform the following steps:

Send a test email.
Check your inbox to confirm that the email has been delivered.

Troubleshooting

If you encounter any issues, please check the following:

Ensure that the Gmail API is correctly enabled.
Verify that the OAuth credentials are valid.
Check the log file for more information.

💻 Usage Examples

Basic Usage

The server provides several tools for email operations. Here are the details of each tool:

Tool	Function	Input	Return
send-email	Send an email to the recipient's address	`recipient_id` (string): Recipient's email address `subject` (string): Email subject `message` (string): Email content	Status and message ID
trash-email	Move an email to the trash	`email_id` (string): Automatically generated email ID	Success message
mark-email-as-read	Mark an email as read	`email_id` (string): Automatically generated email ID	Success message
get-unread-emails	Retrieve unread emails	None	List of emails, including email IDs
read-email	Retrieve the content of a given email	`email_id` (string): Automatically generated email ID	Dictionary of email metadata and mark the email as read
open-email	Open an email in the browser	`email_id` (string): Automatically generated email ID	Success message and open the specified email in the default browser