agent-browser

You can automate browser tasks from the command line by using this Skill's command-based actions like open, snapshot, fill, and click. It handles navigation and data capture for one-shot tasks or scripted flows without requiring full Playwright setup.

Element ref-based interaction uses snapshot-driven element references to minimize context size and improve repeatability. Instead of raw selectors, you interact with identified refs from a page snapshot to fill fields and click elements reliably.

Yes, you can run parallel browser sessions and configure optional viewport or device emulation. The Skill supports session persistence alongside these configurations to facilitate testing and scraping workflows.

Yes, the browser automation tool supports optional proxy configurations. You can pass proxy settings alongside viewport and device emulation parameters when launching commands for navigation and interaction.

The best way to capture a screenshot from the terminal is using the snapshot command after opening a URL. This captures the page state and identifies interactive elements for later automated actions.

You should not use a CLI browser automation tool if your testing requires complex visual regression analysis or extensive continuous integration pipelines. It is designed for one-shot tasks and quick scripted flows rather than full test suite management.