Agent Browser Automation

•
1
•

Automates browser navigation, DOM interactions, visual regression testing, and data extraction through CLI commands.

Agent Browser provides a comprehensive command-line browser automation toolkit tailored for AI agents and developers. It allows Claude Code to open web pages, analyze accessibility trees with interactive element reference IDs, execute clicks, keyboard input, and form fills, handle cookies and local storage, capture full-page and annotated screenshots, record videos, intercept network requests, and perform visual and snapshot diffs. It is ideal for end-to-end web testing, workflow automation, and structured data extraction across complex, dynamic web applications.

主な機能

01Element referencing via accessibility snapshots for precise, deterministic UI interactions
021 GitHub stars
03Annotated screenshots and full-session video recording for visual inspection and debugging
04Visual pixel diffing and accessibility tree comparison against baselines or between URLs
05Network request interception, mocking, route aborting, and API traffic analysis
06Comprehensive session, cookie, and local storage state persistence across runs

ユースケース

01Form filling, UI workflow automation, and generating visual documentation or PDF exports
02Complex web scraping and multi-step data extraction requiring authentication and interactive workflows
03End-to-end automated testing and visual regression verification for web applications