How does the AI know which email to read?

It first finds the message with a tool like list_emails or search_email, which returns each email's ID. read_email then takes that ID — plus the account and mailbox for a fast lookup — and returns the full content.

Does my email get uploaded to the cloud?

No. read_email runs entirely on your Mac through the Mail app. The full message content goes straight to your AI client and is never stored on an LMCP server or any cloud database.

Which email accounts does it support?

Any account configured in Mail on your Mac — iCloud, Gmail, Outlook/Exchange and others. Passing account= (and mailbox= when known) makes the lookup faster and more precise.

Read Full Email Content with AI on Mac

The LMCP read_email tool lets your AI assistant open and read the full body of a specific email straight from Mail on your Mac. Instead of a short subject-line preview, the AI gets the complete message — headers, full text, quoted threads and all — so it can summarize, draft a reply or extract the details you need. You just point it at an email by its ID, and (for a fast lookup) tell it which account and mailbox to check.

A natural prompt looks like this: "Open the latest email from my landlord in my iCloud inbox and tell me the new rent amount and the date it takes effect." The AI finds the message, calls read_email with the right account and mailbox, and answers from the real contents — not a guess.

Which AI agents work?

read_email works with every major AI client that speaks MCP:

Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf and Zed — these desktop clients auto-configure through a local stdio connection the moment LMCP is installed.
ChatGPT, Claude.ai (web), Grok and Perplexity — these web AIs connect through the LMCP Cloud Relay connector, so they reach your Mac securely without any local config file.

One install, 150+ tools, and every one of these assistants can read your mail.

Automation

read_email is the detail step in longer chains. The AI typically calls list_emails or search_email first to find the right message ID, then read_email to pull the full body, then reply_email or send_email to act on it. It can also feed the contents into create_event to put a mentioned meeting on your calendar, into create_reminder for a follow-up, or into a Notes entry to file the key facts. Reading the email is the bridge between finding it and doing something useful with it.

Context

Because LMCP runs natively on your Mac, the AI works against your real Mail accounts — iCloud, Gmail, work Exchange, whatever you have set up. There is no separate copy to keep in sync and no integration to authorize per account. The assistant sees exactly the email you see, in the mailbox where it actually lives, which is why passing account= and mailbox= keeps the lookup fast and accurate.

Productivity

Reading a long thread, scrolling for the one number that matters, and switching back to your AI to draft a response normally costs minutes per email — and a lot of context-switching. With read_email the AI does it in one shot: "summarize this 40-message thread and tell me what's still unresolved" returns an answer in seconds. Across a busy inbox that's an easy half hour reclaimed every day, with fewer missed details.

Privacy & GDPR

Everything happens on your Mac. read_email opens the message locally through Mail and hands the content directly to your AI client — nothing is uploaded to an LMCP server, stored in a cloud database, or retained anywhere. There are no API keys to leak and no third-party inbox sync. This local-only design makes LMCP GDPR-compliant by architecture, since your email never leaves your machine.

Download LMCP and give your AI assistant the ability to read your real email — free, native, and private.