Browser MCP Agent

0•

Empowers AI agents to control a persistent, real-device fingerprinted browser via Model Context Protocol tools without custom automation scripts.

Browser MCP Agent integrates the AntiBrow engine as an MCP server, enabling Claude and other AI agents to launch, navigate, interact with, and extract data from real web pages through direct tool calls. Unlike traditional headless Chromium setups that get easily flagged by bot detection systems, it features kernel-level fingerprint spoofing across Canvas, WebGL, WebGPU, audio, and network TLS layers to present a consistent, legitimate device identity. With persistent profile management, sessions maintain authenticated logins across restarts without writing Playwright or Puppeteer boilerplate, making it ideal for autonomous web workflows, visual agent debugging, and complex web task execution.

Key Features

01Persistent authenticated sessions across browser runs without re-login requirements
020 GitHub stars
03Support for mobile device emulation and background window execution across macOS, Linux, and Windows
04Kernel-level real-device fingerprinting that matches Canvas, WebGL, audio, and TLS signatures
05Complete MCP browsing toolset covering navigation, clicks, typing, content extraction, and screenshots
06Live View streaming for real-time monitoring and debugging of agent browser activity

Use Cases

01Automating ad-hoc administrative dashboard checks and form-filling tasks without writing automation code
02Autonomous agent navigation and data extraction on complex, login-protected web applications
03Computer-use and browser-use workflows requiring realistic hardware fingerprinting over synthetic headless browsers