What problem does it solve?
Provides a CLI-driven browser automation interface that lets AI agents and developers perform navigation, form filling, clicking, screenshots, data extraction, and E2E interactions without writing complex Playwright scripts, reducing manual web interaction and brittle selector work.
Core Features & Use Cases
- Interactive Snapshots & Refs: Compact accessibility-tree snapshots with stable refs (@e1, @e2) for precise element targeting and minimal context usage.
- Session & State Management: Named isolated sessions, save/load state, and encrypted session persistence for authenticating once and reusing across runs.
- Full Automation Surface: Navigation, clicks, typing, file downloads, screenshots, PDF export, visual diffing, recording, CDP connectivity, mobile simulator support, and proxy configuration for scraping and testing.
- Security Controls: Domain allowlists, action policies, content boundaries, and environment-configurable limits to reduce unsafe automation.
- Use Cases: Automated form submission, authenticated scraping, visual regression and diffing, CI test evidence recording, and reproducing user flows for debugging.
Quick Start
Open the target URL, take an interactive snapshot to obtain refs, then instruct the agent to fill the credential ref and click the submit ref to complete the login flow.