Agent Browser Automation

•
9
•

Enables autonomous web interaction and browser automation via a stateful headless CLI and accessibility-first navigation.

The Agent Browser skill empowers Claude to interact with the live web using a stateful CLI interface designed for precise, accessibility-first navigation. By generating structural snapshots of web pages, it allows the agent to identify elements, click buttons, fill forms, capture screenshots, and manage authenticated sessions via cookie injection. This skill is essential for tasks requiring real-time web interaction, dynamic content scraping from JavaScript-heavy sites, or executing multi-step automated workflows within a headless Chromium environment.

主な機能

01Accessibility tree-based navigation for precise element targeting
02Chrome DevTools Protocol (CDP) support for host browser connection
03High-fidelity screen capture and visual verification capabilities
04Stateful browser session management with open, interact, and close loops
059 GitHub stars
06Session persistence through manual cookie and authentication management

ユースケース

01Visual regression testing and UI auditing through automated screenshots
02Automating complex form submissions and multi-page web workflows
03Extracting data from single-page applications (SPAs) that require JS rendering