Chrome Devtools MCP is a server tool by Benjaminr, built with Node.js and distributed as an npm package. It enables AI agents and large language models to interact with live Chrome browser instances. By connecting to Chrome's DevTools automation capabilities, it provides a powerful interface for web application analysis and control. Available via npx or as a .dxt extension for clients like Claude Desktop, it extends browser interaction beyond traditional scripting. This tool is designed to bring advanced browser debugging and automation directly to AI workflows.

What does Chrome Devtools MCP connect to?

The Chrome Devtools MCP tool functions as a Model Context Protocol server. It establishes a connection with a running Chrome browser instance. This connection allows it to utilize Chrome's DevTools automation and debugging features. It achieves this by employing the Chrome DevTools Protocol (CDP) and Puppeteer. This setup enables AI agents to access and control browser functionalities, making the browser a programmable environment for various tasks. The server acts as an intermediary, translating AI agent requests into browser actions and data retrieval.

Core features and tools

Chrome Devtools MCP exposes many DevTools capabilities to AI agents, offering over 26 tools. These tools span categories such as input automation, navigation, debugging, network, performance, and emulation. Specific callable tools include: navigate_page, performance_start_trace, performance_stop_trace, performance_analyze_insight, evaluate_script, click, upload_file, take_snapshot, and get_connection_status(). There is a slight difference in naming for console and network message tools, with some references using list_console_messages and list_network_requests, while others use list-console_messages and listnetworkrequests.

Who is it for?

The primary audience for Chrome Devtools MCP includes AI coding assistants, large language models, and frontend development teams. It serves those who need to automate browser interactions, debug web applications, or analyze performance without direct manual intervention. Confirmed compatible AI clients include Antigravity, Claude, Gemini CLI, JetBrains AI Assistant, and Visual Studio, among many others. This tool is valuable for integrating browser control into automated development pipelines or AI-driven analysis. Skip this if your work does not involve AI agents interacting with a Chrome browser.

How to get started

No credentials are required for Chrome Devtools MCP. The CLI client uses Unix sockets on Linux/Mac and named pipes on Windows for communication with a background daemon.

Use cases

• Automated performance analysis and optimization: AI agents can record performance traces, extract metrics, and generate insights, enabling automated performance audits in CI/CD pipelines.
• In-depth debugging of web applications: AI agents can inspect the DOM and CSS, analyze network requests, read console output, execute JavaScript, and diagnose issues like CORS errors.
• Reliable browser automation and interaction: Agents can autonomously navigate pages, click buttons, fill forms, take screenshots, and simulate user behavior without direct Puppeteer scripting.

Chrome Devtools MCP provides a direct bridge between AI agents and Chrome's DevTools, enabling advanced browser control and data extraction. Its capabilities support automated web application analysis, debugging, and interaction, making it a valuable asset for development workflows. The tool simplifies complex browser operations for AI systems, opening possibilities for new automation and testing paradigms. Try automating a performance trace and insight generation using performance_start_trace and performance_analyze_insight.