EmailAssistant

EmailAssistant MCP Server enables Claude Desktop to securely access your Gmail account via the Gmail API, fetch emails, and return concise summaries on demand. It lets you query recent activity or specify criteria to get exactly the emails you care about, all through natural, hands-free interactions with Claude.

How to use

You interact with EmailAssistant through Claude Desktop by calling its exposed functions. After you set up the server, Claude will recognize the EmailAssistant MCP server and you can ask for email insights in plain English. Useful workflows include getting a recent summary of messages, retrieving top matching emails based on keywords, or asking for a quick briefing on inbox activity.

How to install

Prerequisites you need on your machine before installing EmailAssistant MCP Server:

• Python 3.9 or newer

• uvx (or use uv) to run the local MCP server

• Python packages: google-api-python-client, google-auth-httplib2, google-auth-oauthlib, mcp[cli]

Follow these steps to set up EmailAssistant MCP Server locally:

Install and configure EmailAssistant MCP Server

# 1) Prepare Gmail API credentials in Google Cloud Console
#    - Create a project or choose an existing one
#    - Enable Gmail API under APIs & Services
#    - Create OAuth 2.0 Client ID credentials
#    - Download the client_secret file (e.g., client_secret_XXXXX.json)

# 2) Create your Python config file
#    This points to your credentials and token storage

import os

SCOPES = ["https://www.googleapis.com/auth/gmail.readonly"]
PROJECT_DIR = "/absolute/path/to/your/project/directory/"
CREDENTIALS_FP = os.path.join(PROJECT_DIR, "client_secret_XXXXX.json")
TOKEN_FP = os.path.join(PROJECT_DIR, "token.json")

# 3) Claude Desktop configuration to run the local MCP server

{ "mcpServers": { "EmailAssistant": { "command": "uv", "args": [ "run", "--with", "mcp[cli]", "--with", "yake", "--with", "google-api-python-client", "--with", "google-auth-httplib2", "--with", "google-auth-oauthlib", "mcp", "run", "/absolute/path/to/your/project/directory/main.py" ] } } }

- Ensure your absolute path to the project directory and main.py matches your setup.
- Install required Python packages as shown in the prerequisites to satisfy the dependencies when the MCP server starts.
- If you run into credential prompts, complete the OAuth flow in a browser to authorize Gmail access for the client secret file you downloaded.

Available tools

get_email_summary

Fetches a summary of emails based on provided time filters or query criteria, returning a concise briefing of unread or newer messages.

get_top_matching_email

Retrieves the top email that best matches given keywords or query terms, useful for quickly spotting important messages.