Chrome DevTools for Agents
Lets coding agents control and inspect a live Chrome browser with DevTools performance insights, network debugging, source-mapped console errors, and Puppeteer automation.
About this Server
Chrome DevTools for Agents (chrome-devtools-mcp) is the official Google Chrome DevTools Model Context Protocol server and CLI tool. It gives AI coding assistants (including Antigravity, Claude, Cursor, VS Code, and Copilot) access to the full power of Chrome DevTools. Key capabilities include recording traces to extract actionable performance insights and Core Web Vitals, in-depth network request and header analysis, full DOM accessibility snapshots and screenshots, live console message tracking with source-mapped stack traces, and deterministic Puppeteer automation that automatically awaits action results.
Direct low-latency RPC protocol communication with structured JSON schemas.
Restricted capability bounds to prevent accidental modifications or leakage.
Security & Best Practices
When to Use (and When Not To)
- •In-depth browser performance audits, recording DevTools traces, and optimizing Web Vitals (LCP, CLS, INP)
- •Inspecting HTTP network requests, headers, and failed API payloads from local development servers
- •Debugging client-side JavaScript runtime errors and exceptions with source-mapped stack traces
- •Automating browser interactions and form flows deterministically using official Puppeteer drivers
- •Memory leak debugging with heap snapshots and accessibility (a11y) tree verification
- •Headless cloud pipelines lacking Chrome binaries or display server environments without --headless flag
- •Large-scale web scraping across millions of pages (use crawling tools or specialized APIs)
- •Direct database manipulation or SQL queries (use PostgreSQL, Supabase, or Neon MCP instead)
Install & Setup
claude mcp add chrome-devtools --scope user npx -y chrome-devtools-mcp@latestExposed Tools & Endpoints
12 toolsFunctions your AI agent can invoke via this Model Context Protocol server
Starts recording a Chrome DevTools performance trace to capture timeline events and CPU activity.
Stops the active performance trace recording and compiles performance data.
Extracts actionable DevTools performance insights, Web Vitals metrics (LCP, CLS, INP), and rendering bottlenecks.
Retrieves browser console log messages, uncaught exceptions, and warnings with source-mapped stack traces.
Inspects network requests, response payloads, HTTP headers, status codes, and timing waterfall charts.
Captures a viewport or full-page screenshot of the active browser tab for visual verification.
Captures a structured DOM snapshot or accessibility tree representation for semantic reasoning.
Evaluates JavaScript expressions within the execution context of the active page or service worker.
Clicks an interactive DOM element identified by selector or element UID using Puppeteer automation.
Batch-fills form fields (text inputs, checkboxes, select dropdowns) in a single deterministic step.
Navigates the browser page to a specified URL, reloads, or traverses browser history.
Opens a new browser tab and navigates directly to a target URL or local development server.