Back to browse
Local MCP – Claude/ChatGPT read your iMessage, Teams, files on-device

Local MCP – Claude/ChatGPT read your iMessage, Teams, files on-device

by lanchuske·Jul 5, 2026·2 points·0 comments

Post Description

I kept hitting the same wall: AI assistants are great until they need my context — the email thread, the calendar, the iMessage, the file on my disk. Cloud connectors can only reach things with a public API, so they physically can't touch the stuff that lives only on my machine.

So I built Local MCP — a native macOS app that connects Claude Desktop, Claude Code, Cursor, and ChatGPT to your real local apps over MCP: Mail, Calendar, Contacts, iMessage, Microsoft Teams, Slack, WhatsApp, Notes, Reminders, OmniFocus, OneDrive/Google Drive, Office docs, and local files. ~180 tools.

The part I care about most is where the data goes:

- Desktop/CLI clients (Claude Desktop, Cursor, Claude Code): everything runs on localhost — nothing leaves your Mac, no API keys, no OAuth to us.

- Web AIs (ChatGPT, Claude.ai) that can't reach localhost: there's an optional, opt-in encrypted relay you turn on yourself. It's off by default and it's the only path anything leaves the machine. I flag this explicitly because "100% local" gets thrown around loosely.

How it reaches the hard stuff: Calendar/Contacts via EventKit; Mail/Notes/Messages via AppleScript/JXA against the running apps; Teams/Slack/WhatsApp by reading their local on-disk stores (IndexedDB/LevelDB) rather than any Graph API — so it sees history a cloud integration can't. Destructive actions (send email, delete event) always preview and ask first.

It's past the experiment stage — a few hundred people run it and the heavy users do thousands of operations a day. It's free, no paid tier yet. macOS-only for now (closed-source today).

Install: signed .app at https://local-mcp.com, or `curl -fsSL 'https://local-mcp.com/install?ref=show-hn' | bash`

Happy to go deep on anything — the local-store readers, the JXA/TCC permission model, the opt-in relay design, why macOS-only.

Similar Projects