agent-browser

Automate browser interactions by issuing sequential commands to navigate URLs, fill forms, click elements, and extract data. This enables end-to-end automation for tasks like form submission, authentication flows, and data collection within single or multi-step sessions.

Data extraction from web pages is handled by navigating to the target URL, taking a DOM snapshot to identify elements, and using text extraction commands. You can also capture visual reports via screenshots and PDF output during the automation session.

Yes, a local Chromium-based browser is required to execute web automation commands. The tool relies on this environment to maintain per-session state and perform interactions like opening pages, filling forms, and taking screenshots.

Yes, authentication flows and form filling are supported in multi-step sessions. The command surface includes open, fill, click, and wait operations, allowing you to manage per-session state and complete complex end-to-end automation sequences.

Web automation testing supports visual validation through screenshot and PDF output commands. You can capture the visual state of pages during single or multi-step sessions to verify rendering and interactive elements.

A DOM snapshot is needed to identify interactive elements before clicking. By capturing the page structure first, the automation tool can accurately target elements for subsequent fill, select, and type commands in the workflow.