Model Context Protocol (MCP) server that bridges your personal WhatsApp account to Claude / Cursor via a Go-based WhatsApp bridge and a Python MCP server.
https://github.com/lharries/whatsapp-mcpFinally, a proper bridge between your WhatsApp conversations and Claude. This MCP server connects your personal WhatsApp account directly to Claude Desktop or Cursor, giving you AI-powered message analysis, smart replies, and automated workflows using your actual chat history.
Your WhatsApp contains years of conversations, contacts, shared files, and context that could be incredibly valuable for AI assistance. But it's trapped in a mobile app with no API access for personal accounts. When you need Claude to help with something related to your messages - analyzing conversation patterns, drafting replies based on context, or finding that document someone shared months ago - you're stuck copying and pasting.
This MCP server solves it by connecting directly to WhatsApp Web's multidevice API using the solid whatsmeow library. No third-party services, no data sharing - everything stays on your machine.
Two-component architecture:
Your messages get indexed locally, and Claude only sees what you explicitly ask it to access through the available tools.
Message Intelligence:
Smart Communication:
Media Management:
Contact Organization:
Conversation Analysis:
"Analyze my last 50 messages with John and suggest three follow-up topics"
Claude searches your message history, identifies conversation threads, and suggests contextually relevant follow-ups.
Smart Group Management:
"Check what the family group has been discussing this week and send an update about my travel plans"
Claude reviews recent group activity and crafts an appropriate response matching the conversation tone.
Media Workflow:
"Find that contract PDF Mike sent last month and summarize the key terms"
Claude searches message history, downloads the file, and provides analysis.
Prerequisites:
Authentication is one QR scan:
cd whatsapp-bridge && go run main.go
Scan the QR code with WhatsApp mobile, and you're connected. The bridge maintains the session and handles reconnections.
MCP integration through standard config:
{
"mcpServers": {
"whatsapp": {
"command": "/path/to/uv",
"args": ["--directory", "/path/to/whatsapp-mcp-server", "run", "main.py"]
}
}
}
All message data stays on your machine in SQLite. The Go bridge only syncs what's necessary, and the MCP server only shares specific messages when you use tools that require them. No cloud services, no data mining - just your personal WhatsApp integrated into your AI workflow.
You control exactly what Claude accesses through explicit tool usage. Want to search messages? Use the search tool. Need to send something? Use the send tool. Claude can't browse your messages without you initiating the action.
Windows compatibility requires CGO enabled and a C compiler (MSYS2 recommended) since the SQLite driver needs native compilation.
Session management handles authentication renewal automatically, though you might need to re-scan the QR code every ~20 days.
Media handling includes automatic audio conversion to WhatsApp's Opus format when FFmpeg is available, or raw file sending as fallback.
This bridges a real gap in personal productivity workflows. Instead of context-switching between WhatsApp and Claude, you now have your entire message history available as AI context when you need it.
The 4,200+ stars suggest developers have been waiting for exactly this kind of integration. Your WhatsApp conversations can finally contribute to your AI-enhanced workflows instead of living in isolation.