Apple Mail Summary MCP Server

A Model Context Protocol (MCP) server for macOS that allows AI agents to interact with the local Apple Mail application. This server enables LLMs to fetch emails from specific accounts and mailboxes, and includes specialized tools for parsing academic paper alerts.

Features

• 📧 Fetch Emails: Retrieve the latest unread emails from any configured Apple Mail account and mailbox.

• 🎓 Scholar Tools: specialized parsing for Google Scholar Alert emails to extract paper titles and links.

• 🔒 Local & Secure: Runs locally on your machine, communicating directly with Apple Mail via AppleScript.

Related MCP server: mcp-imap-server

Prerequisites

• macOS: This server relies on osascript (AppleScript) to communicate with Apple Mail. It will not work on Windows or Linux.

• Apple Mail: Must be set up and running with your email accounts configured.

• Python 3.10+

• uv (recommended for package management) or standard pip.

Usage

1. Identify your Mail Configuration

To use this server, you need to know the exact names of your Accounts and Mailboxes as they appear in Apple Mail.

You can find these by hovering over the mailbox in the Mail app sidebar, or by running this simple AppleScript in 'Script Editor.app':

tell application "Mail"
    get name of every account
end tell

2. Environment Variables (Optional)

You can set default values for the account and mailbox using environment variables. This allows the AI agent to skip specifying these parameters every time.

3. Configuration for Claude Desktop

You can add this server to your claude_desktop_config.json in two ways:

Option A: Running directly from GitHub via uvx (No manual install needed)

This is the easiest way to try it out without manually cloning the repo.

{
  "mcpServers": {
    "apple-mail-summary": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/BillShiyaoZhang/apple-mail-summary-mcp.git", 
        "apple-mail-summary-mcp" 
      ],
      "env": {
        "APPLE_MAIL_SUMMARY_MCP_ACCOUNT": "Your Default Account Name",
        "APPLE_MAIL_SUMMARY_MCP_MAILBOX": "Inbox"
      }
    }
  }
}

Option B: Running from a Local Clone (Development)

If you want to run from a local copy or make changes, first clone the repository and install dependencies:

Using uv (Recommended)

git clone https://github.com/BillShiyaoZhang/apple-mail-summary-mcp.git
cd apple-mail-summary-mcp
uv pip install -e .

Using pip

git clone https://github.com/BillShiyaoZhang/apple-mail-summary-mcp.git
cd apple-mail-summary-mcp
pip install -e .

Then configure your claude_desktop_config.json:

{
  "mcpServers": {
    "apple-mail-summary": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/apple-mail-summary-mcp",
        "run",
        "apple-mail-summary-mcp"
      ]
    }
  }
}

Note: For Option B to work, make sure uv is in your system PATH.

3. Running the Server Manually

You can also run the server directly if needed:

python -m apple_mail_summary_mcp.server

4. Available Tools

fetch_emails(account: str = None, mailbox: str = None, limit: int = 5)

Fetches a list of unread emails from a specific source.

• account: The exact name of the account (e.g., "iCloud", "Exchange", "Personal"). Defaults to APPLE_MAIL_SUMMARY_MCP_ACCOUNT if not provided.

• mailbox: The mailbox name (e.g., "Inbox", "Spam", "Google Scholar"). Defaults to APPLE_MAIL_SUMMARY_MCP_MAILBOX (or "Inbox") if not provided.

• limit: Maximum number of emails to retrieve.

extract_scholar_links(html_content: str)

A utility tool to parse the HTML content of a Google Scholar Alert email and return structured data (Title, URL).

• html_content: The raw HTML body of the email.

Development

Project structure:

• src/apple_mail_summary_mcp/email_client.py: Handles AppleScript communication.

• src/apple_mail_summary_mcp/processor.py: Contains HTML parsing logic (BeautifulSoup).

• src/apple_mail_summary_mcp/server.py: Defines the FastMCP server and tools.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT