Browser Tools MCP, developed by Agentdeskai, is an MCP server designed to empower AI agents with real-time access to browser data. Built with Node.js, it connects directly to a real Chrome or Chromium-based browser session through its DevTools interface. This tool enables AI clients to monitor and interact with web pages, capturing crucial information for various development workflows. It integrates with several compatible AI clients, providing a critical bridge between AI applications and live browser environments. Developers can utilize its capabilities to enhance debugging, auditing, and optimization tasks within their AI-driven development processes.

What does Browser Tools MCP connect to?

Browser Tools MCP establishes a connection with a real Chrome or any Chromium-based browser session. It specifically interfaces through the browser's DevTools, enabling deep monitoring and interaction. This connection allows the MCP server to function as a vital link for AI-powered applications, providing them with direct access to a live web environment. Its primary role is to serve as a browser monitoring and interaction tool, translating real-time browser states into actionable data for AI agents.

Core features and tools

The tool offers extensive functionality for capturing various browser activities. It can retrieve console logs, monitor network activity, extract DOM elements, and capture full-page screenshots. Furthermore, Browser Tools MCP runs Lighthouse audits covering accessibility, performance, SEO, and best practices. While the system supports 16 tools, specific shipped prompts include debuggerMode, auditMode, and nextjsSeoAudit.

Who is it for?

Browser Tools MCP targets developers who utilize MCP-compatible AI clients and require their AI agents to access real-time browser data. Its design caters to those integrating AI into their development workflows for tasks like debugging and auditing web applications. Compatible AI clients include Cursor, Claude Code, Windsurf, Cline, Zed, Gemini CLI, RooCode, and Claude Desktop. Skip this if your development workflow does not involve MCP-compatible AI clients interacting with browser sessions.

How to get started

Version 2.0 of the server authenticates using a per-run bearer token, with credentials scrubbed before leaving the browser. The MCP server itself supports "stdio" as a transport protocol.

Use cases

• Debugging browser console errors directly within an AI client, eliminating the need for context switching during development.
• Automating Web Content Accessibility Guidelines (WCAG) compliance checks throughout the development process.
• Improving on-page Search Engine Optimization (SEO) by initiating Lighthouse-based audits from an AI chat interface.

Browser Tools MCP offers a focused solution for integrating AI agents with real-time browser interactions. Its ability to capture logs, network data, and run Lighthouse audits directly from an AI client provides significant utility for developers. The tool supports multiple AI clients and uses a secure, token-based authentication model. It enables AI to assist with critical web development tasks, from debugging to compliance.