What problem does it solve?
Manual browser testing is slow, error-prone, and impossible to scale across deployments and devices. AI agents need reliable programmatic browser control to verify sites, test user flows, and capture visual evidence without manual clicking or flaky human observation.
Core Features & Use Cases
- Persistent Headless Browser: Auto-starts a Chromium daemon in ~3 seconds, then executes commands in ~100-200ms. Cookies, tabs, and login sessions persist between calls for seamless multi-step testing.
- Element Interaction & Verification: Navigate pages, click/fill/select elements using accessibility refs, verify element states (visible, enabled, checked), and diff page changes after interactions to confirm they worked.
- Visual Bug Capture: Take screenshots in multiple modes (full page, viewport, element crop, region clip, base64), generate annotated screenshots with ref labels, and test responsive layouts across mobile, tablet, and desktop viewports.
- Form & Upload Testing: Test file uploads, fill complex forms, handle dialogs with auto-accept/dismiss, and capture console errors and network requests to catch silent failures.
- Use Case: A developer deploying a new feature can automatically verify the staging URL loads correctly, screenshots render properly on all device sizes, the login flow completes without errors, and no console warnings appear — all in seconds without manual clicking.
Quick Start
Navigate to your staging URL and take a screenshot to verify the deployment looks correct.