Overview
Hand a real Chromium to the Agent — navigate, screenshot, fill forms, click, run scripts. Useful for testing, scraping structured data, and automating sign-ins.
Features
- Real browser + accessibility tree
- Screenshots, PDFs, network capture
- Element waits and timeouts
- Multi-tab and session support
- Record into replayable scripts
Exposed tools
Once connected, the Agent can call these tools directly in conversation
navigateclickfillscreenshotsnapshotevalnetwork_logs
