Claude Chrome MCP
Enables MCP clients like Claude Desktop to automate and interact with claude.ai via a Chrome extension and WebSocket relay.
README
Claude Chrome MCP
Browser automation tool enabling MCP clients (Claude Desktop, Claude Code, Cursor) to interact with claude.ai through Chrome extension and MCP server.
Components
- MCP Server
- Chrome Extension (enabled only for claude.ai)
- CLI Tool
Quick Start
Prerequisites
- Google Chrome browser
- Node.js v16 or higher
- npm installed
Step 1: Install Chrome Extension
- Open Chrome and navigate to chrome://extensions/
- Enable "Developer mode" (toggle in top right)
- Click "Load unpacked" and select the
extension/directory from this project - The extension icon should appear in your toolbar
Step 2: Install Dependencies
# From the project root directory
cd mcp-server
npm install
Step 3: Configure Claude Desktop
- Open Claude Desktop settings
- Navigate to Developer > Edit Config
- Add this configuration:
{
"mcpServers": {
"claude-chrome-mcp": {
"command": "node",
"args": ["<absolute-path-to-project>/mcp-server/src/server.js"]
}
}
}
Replace <absolute-path-to-project> with the full path to this project (e.g., /Users/yourname/claude-chrome-mcp)
Step 4: Restart Claude Desktop
After adding the configuration, restart Claude Desktop to load the MCP server. The server will automatically start its embedded WebSocket relay on port 54321.
Step 5: Test the Connection
In Claude Desktop, you can now use natural language commands like:
- "Check Chrome status"
- "Open a new tab"
- "Send 'Hello, World!' to the new Claude tab"
- "List all open Claude tabs"
- "Search for Claude chats with 'math'"
Key Features
- Async message sending with completion detection
- Claude-to-Claude response forwarding
- Conversation management via Claude.ai API
- Network inspection and debugging tools
- WebSocket-only architecture with persistent connections
- Health monitoring endpoint for relay status
Documentation
| Guide | Description |
|---|---|
| Architecture Analysis | Critical issues and improvement roadmap |
| Architecture | System design and components |
| Troubleshooting | Complete troubleshooting and timeout resolution |
| Development Workflows | Code changes, testing, and development procedures |
| TypeScript | Type definitions and development |
| CLI Tools | Command-line interface and setup |
| Event Architecture | Visual system diagrams |
Project Resources
- MCP Configuration - Quick commands and setup
- GitHub Releases - Release history
- GitHub Issue Script - Claude Code integration utilities
CLI Tool (Optional)
npm install && npm run build && npm link
Architecture
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Claude.ai │ │ Chrome Extension │ │ WebSocket Relay │
│ (Browser) │◄──►┤ (Offscreen Doc) │◄──►│ Port 54321 │
└─────────────────┘ └──────────────────┘ └─────────────────┘
▲
│
┌────────┴────────┐
│ │
┌───────▼──────┐ ┌───────▼──────┐
│ MCP Server │ │ Claude Code │
│ (Relay Mode) │ │ MCP Client │
└──────────────┘ └──────────────┘
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。
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。
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。
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。
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。
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。
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。