agent-browser
Headless browser automation CLI optimized for AI agents with accessibility tree snapshots and ref-based element selection.
Overview
Agent Browser is a headless browser automation CLI optimized for AI agents. It uses accessibility tree snapshots with refs for deterministic element selection, making it ideal for automating multi-step workflows, especially when performance is critical.
Key Features
- Fast browser automation using accessibility tree snapshots
- Deterministic element selection with refs
- Session isolation for isolated browsers
- State persistence for cookies and storage
- Network control for blocking or mocking requests
How It Works
Agent Browser uses a core workflow that involves navigating and snapshotting, parsing refs from JSON, and interacting with elements. It also provides various commands for navigation, snapshotting, interactions, getting information, checking state, waiting, sessions, state persistence, screenshots, PDFs, network control, cookies, storage, tabs, and frames.
Use Cases
- Automating multi-step workflows
- Deterministic element selection for complex SPAs
- Performance-critical applications
- Session isolation for isolated browsers
- State persistence for cookies and storage
- Network control for blocking or mocking requests
Anmeldelser
Ingen anmeldelser ennå.