Agent Browser Automation
Automates Chrome Canary for web testing, data extraction, and interactive page manipulation using Claude Code.
Agent Browser provides a powerful interface for Claude Code to interact directly with the web using Chrome Canary. It enables complex workflows like automated form submission, visual testing via screenshots, and semantic data extraction through accessibility tree analysis. By utilizing a unique reference system for page elements, it allows for high-precision interactions, session management, and the ability to save or load authentication states, making it an essential tool for developers building agents that need to navigate and process live web environments.
Características Principales
01Interactive element identification using unique reference tags (@e1, @e2) from snapshots.
02Capture high-quality screenshots and page snapshots for visual verification and debugging.
03Full browser control including navigation, clicking, typing, hovering, and scrolling.
0415 GitHub stars
05Semantic locators and accessibility tree parsing for robust element finding and data extraction.
06Session and state management for persisting authentication, cookies, and local storage.
Casos de Uso
01Automating repetitive web-based administrative tasks and complex form submissions.
02Automated End-to-End (E2E) testing for web applications directly from the terminal.
03Web scraping and structured data extraction from dynamic, JavaScript-heavy websites.