Agent Browser Automation

•
707
•

Automates Chrome and Chromium interactions using a high-performance, token-efficient CLI designed for AI agents.

The agent-browser skill empowers Claude to perform complex web interactions through a lightweight, CDP-based automation engine. By replacing heavy HTML parsing with accessibility-tree snapshots and compact '@eN' references, it allows agents to navigate, interact with elements, and extract data using significantly fewer tokens than traditional methods. It supports advanced features like session persistence, multi-tab management, network mocking, and annotated screenshots, making it the ideal solution for web testing, automated data collection, and real-time browser-based workflows.

主要功能

01Network mocking and request routing for robust web testing and scraping
02Session persistence and multi-tab management for complex user workflows
03Full browser interaction suite including clicks, forms, uploads, and drag-and-drop
04Advanced visual capabilities with annotated screenshots and video recording
05707 GitHub stars
06Token-efficient page snapshots using accessibility-tree (@eN) references

使用场景

01Structured data extraction and web scraping from dynamic JavaScript sites
02Automated end-to-end testing of web applications directly within Claude
03Automating repetitive web tasks such as form submissions and account management