Give your AI coding assistants (Claude, Cursor, Copilot) full access to Chrome DevTools - they can debug, profile performance, and automate browsers like a human developer would.

Ever wished your AI coding assistant could actually open a browser, inspect network requests, and debug JavaScript like you do? That’s exactly what this MCP server delivers. It bridges the gap between AI agents and Chrome DevTools, letting tools like Claude or Cursor control a live Chrome instance with full debugging capabilities.

Built on Puppeteer and the Model Context Protocol, it goes far beyond basic browser automation. Your AI can capture performance traces, analyze Core Web Vitals, screenshot visual regressions, and even inspect source-mapped stack traces from console errors. The 23k+ stars reflect its solid execution - Google maintains this with proper error handling, telemetry, and integration with their CrUX performance database for real-world metrics.

Perfect for teams building AI-powered testing workflows, automated performance audits, or debugging agents. Setup is straightforward with Node 20+ and works with any MCP-compatible AI tool. The comprehensive tool reference and troubleshooting docs show this isn’t a weekend hack - it’s production-ready infrastructure for the AI-assisted development workflow.

⭐ Stars: 23944
💻 Language: TypeScript
🔗 Repository: ChromeDevTools/chrome-devtools-mcp