Google launches Chrome DevTools MCP allowing AI agents to debug, automate & test in Chrome

Google has released the public preview of Chrome DevTools MCP, a Model Context Protocol server that allows AI coding agents to connect with a live Google Chrome browser. This new tool enables agents such as Gemini CLI, Claude Code, Cursor, and GitHub Copilot to access Chrome’s debugging surface, bridging the gap between code generation and runtime testing. By observing live browser behavior, AI agents can now validate, test, and debug suggested changes directly in real time, moving beyond static code suggestions.

With Chrome DevTools MCP, supported actions for AI assistants include recording performance traces, inspecting Document Object Model and CSS, running JavaScript, reading console logs, and automating user flows. Additional operations cover navigation, simulating clicks and user input, evaluating scripts, listing network requests or console messages, and retrieving screenshots or DOM snapshots. These capabilities make it possible to automate audits, reproduce bugs, test fixes, and diagnose layout or network issues directly from AI workflows.

Under the hood, puppeteer manages browser automation, and communication happens through the Chrome DevTools Protocol. Installation is reduced to a single configuration that runs npx chrome-devtools-mcp@latest, provided Node.js 22 and a current Chrome version are available. Google also offers setup commands tailored to various coding environments to speed adoption.