Agent Browser Automation
Automates headless browser workflows using accessibility tree snapshots and deterministic element selection for AI agents.
Agent Browser is a specialized skill designed to empower AI agents with robust, high-performance web automation capabilities. Unlike standard visual-based tools, it leverages accessibility trees and reference-based selection to ensure deterministic interactions with complex Single Page Applications (SPAs). It supports advanced features such as session isolation, authentication state persistence, and network request mocking, making it ideal for multi-step workflows, web scraping, and automated testing where reliability and speed are paramount.
Key Features
01Multi-session support for managing isolated browser contexts simultaneously
02Advanced network control for blocking requests or mocking API responses
03Deterministic element selection using accessibility tree snapshots and @refs
04Authentication state persistence to save and load cookies or local storage
05Comprehensive interaction suite including click, fill, hover, scroll, and drag
063 GitHub stars
Use Cases
01Automating complex, multi-step user flows in modern web applications
02Performing structured web scraping with precise, reference-based data extraction
03Conducting automated end-to-end testing across different user roles and sessions