Do I need a WhatsApp API key or Business account?

No. LMCP reads WhatsApp locally on your Mac through Wacli, an unofficial bridge to your existing WhatsApp session. There are no API keys, tokens, or Business account requirements.

How does the AI know which chat to read?

It first calls whatsapp_list_chats to list your conversations and pick the right chat_id, then passes that id into whatsapp_read_messages. You just describe the chat in plain language and the AI resolves it.

Are my WhatsApp messages sent to a server or to the AI company?

No. The messages are read on your Mac and only the assistant's request and its answer move. Nothing is stored in the cloud, keeping the workflow private and GDPR-compliant by architecture.

Read WhatsApp Messages with AI on Mac

The whatsapp_read_messages tool lets your AI assistant pull the actual message history from a specific WhatsApp chat on your Mac. Once you've identified a conversation, the AI reads its messages so it can summarize, search, or act on them. A natural prompt looks like: "Read my latest WhatsApp messages with the Marketing group and tell me if anyone asked me a question I haven't answered." The AI first finds the chat, then reads the thread and reports back — no copy-pasting, no scrolling.

Under the hood the chat_id must come from a previous whatsapp_list_chats call, so the AI knows exactly which conversation to open. LMCP uses Wacli (an unofficial WhatsApp bridge) running locally on your Mac to fetch the messages.

Which AI agents work?

LMCP exposes 150+ tools to any MCP-compatible assistant:

Desktop clients — Claude Desktop, Cursor, VS Code (GitHub Copilot), Windsurf and Zed auto-configure through a local stdio connection. Install LMCP, restart the client, and the tools appear.
Web AIs — Claude.ai (web), ChatGPT, Grok and Perplexity connect through the LMCP Cloud Relay connector, which securely bridges the web assistant to the server running on your Mac.

Either way, the WhatsApp data never leaves your machine — only the assistant's request and the answer travel.

Automation

This tool shines when chained with others. The AI calls whatsapp_list_chats to discover conversations, passes a chat_id into whatsapp_read_messages to read the thread, then routes the result onward — drafting a reply with whatsapp_send_message, creating a follow-up with create_reminder, scheduling a call via create_calendar_event, or emailing a summary with send_email. One prompt can turn "catch me up on this chat" into a fully triaged to-do list.

Context

Because the AI reads your real WhatsApp messages — not a sample or a guess — its answers reflect what people actually said. It can cross-reference a chat against your Calendar, Contacts, Mail or Notes, all living on the same Mac. Ask "who in this WhatsApp thread is also on tomorrow's calendar invite?" and the assistant works across apps with genuine, current data.

Productivity

Catching up on a busy group chat by hand means opening WhatsApp, scrolling through dozens of messages, and trying to remember what needed a response. With whatsapp_read_messages the AI does that in seconds: it summarizes the thread, flags unanswered questions, and surfaces action items. A five-minute scroll becomes a one-line answer, and nothing important slips through.

Privacy & GDPR

Everything runs locally on your Mac. LMCP reads WhatsApp through Wacli on-device — there are no API keys, no cloud copy of your messages, and nothing is stored on a server. Your conversations stay yours. This is privacy by design: GDPR-compliant by architecture. Download LMCP and connect your favorite AI in minutes.