Agent Browser Automation

•
2
•

Automates browser navigation, DOM interactions, form filling, screenshots, and network traffic for web testing and data extraction.

Agent Browser provides Claude with direct control over headless and headed web browsers via an intuitive command-line interface. By generating accessibility snapshots with stable element references, it enables reliable page navigation, clicking, form submission, and data extraction without fragile selectors. The skill also supports advanced browser operations including cookie and session persistence, network request interception and mocking, multi-tab workflows, video recording, and CDP debugging, making it an essential tool for automated testing, synthetic monitoring, and dynamic web scraping.

主な機能

01Session isolation and authentication state persistence across parallel browser instances
02Multimedia capture including full-page screenshots, PDF generation, traces, and WebM video recording
03Network traffic inspection, request mocking, route aborting, and header customization
04Comprehensive user emulation including typing, clicks, drag-and-drop, file uploads, and scrolling
052 GitHub stars
06Element interaction via compact accessibility snapshot references (@e1, @e2) and semantic locators

ユースケース

01Executing automated end-to-end web testing and verifying interactive UI workflows
02Automating multi-step web tasks such as authentication flows, report generation, and form submissions
03Extracting dynamic content and structured data from JavaScript-heavy websites