agent-browser

Automate web scraping and form filling on dynamic websites by using an AI-driven browser to navigate pages, discover UI elements via snapshot refs, and extract results. The browser handles dynamic DOM changes by re-snapshotting after navigation or updates to ensure accurate element targeting.

Browser automation handles authentication and session persistence by maintaining independent named sessions and reusing saved auth state. This approach avoids repeated logins by preserving cookies and session data across multiple automation runs, ensuring continuous access to authenticated areas without manual re-entry.

Yes, you can use semantic locators for reliable UI element targeting during browser automation. The system supports addressing elements using role, text, label, or testid attributes, alongside snapshot-generated refs like @e1, to interact accurately with dynamic web pages and complex UI structures.

Browser automation does work with iOS simulator contexts for web scraping and interactive tasks. The system is designed to handle interactive website automation across both desktop and iOS simulator environments, allowing you to navigate pages, fill forms, and extract results consistently.

The best way to capture evidence of web automation tasks is by taking screenshots or generating PDFs of the page state. After navigating pages and performing actions like form submission, the automation tool captures the final state for review and verification.

Web scraping fails when DOM elements change after page navigation because old element references become stale. To prevent this, you must re-snapshot the page after any navigation or dynamic DOM updates to refresh element references and ensure deterministic waits before extracting content.