whatsapp-mcp-assistant
MCP server that exposes a real WhatsApp account to Cursor through safe, approval-gated tools and a local Go bridge, enabling chat management, contact search, and message send with explicit confirmation.
README
WhatsApp MCP Assistant
whatsapp-mcp-assistant — MCP server that exposes a real WhatsApp account to Cursor through safe, approval-gated tools and a local bridge.
This project exposes a real WhatsApp account to Cursor through an MCP server. Cursor is the agent interface; this repo only provides MCP tools and a local WhatsApp bridge.
V1 uses:
- Python
FastMCPserver for Cursor. - Local Go
whatsmeowHTTP bridge for WhatsApp Web login and messaging. - Real per-user WhatsApp sessions stored locally under
sessions/. - No mock WhatsApp data.
Tools
get_connection_status()get_login_qr()logout_whatsapp(confirm)get_recent_chats(limit=10, unread_only=false, include_groups=true)search_contact(query, include_groups=true, limit=5)get_messages_from_contact(chat_id, limit=20, include_outgoing=true)draft_reply(chat_id, instruction, recent_messages=[], tone="natural")send_whatsapp_message(chat_id, message, approved)
send_whatsapp_message rejects every request unless approved=true.
Setup
Install Python dependencies:
python -m pip install -r requirements.txt
Install Go, then install Go bridge dependencies:
cd go-whatsapp-bridge
go mod tidy
Start the WhatsApp bridge:
go run .
Start Cursor with this MCP server config shape:
{
"mcpServers": {
"whatsapp-assistant": {
"command": "python",
"args": complete path like ["/AI/MCP Server/server.py"]
}
}
}
QR Login Flow
- Start the Go bridge.
- Ask Cursor:
Check WhatsApp connection. - If disconnected, ask Cursor:
Get WhatsApp login QR. - Scan the returned QR string using WhatsApp > Linked Devices.
- Ask Cursor to check connection again.
Every user must run the server locally and scan their own WhatsApp QR. Do not share sessions/, .env, QR codes, tokens, or connected server instances.
Demo Flow
Use a test WhatsApp number.
- Send a WhatsApp message from another phone to the connected test account.
- Ask Cursor:
Show my recent WhatsApp chats. - Ask Cursor:
What did Bilal message me today? - Ask Cursor:
Draft a reply saying I will come at 6. - After reviewing the draft, say:
Okay, send it.
Current Bridge Limitation
The local bridge records real messages it receives while the bridge is running and keeps them in sessions/message_cache.json. WhatsApp Web libraries do not reliably provide full historical chat export through this simple bridge, so test by sending messages after the bridge is connected.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。