Browser Automation

Problem

Web workflows often involve navigation, clicks, form filling, dynamic loading, and result extraction. Doing this manually is repetitive and hard to reuse, while maintaining Python or JavaScript scripts requires managing selectors, retries, proxies, and error states. This skill describes tasks in natural language and drives a headless browser through CLI commands.

How It Works

• Session setup: start a headless browser instance and configure proxy and User-Agent values.
• Page actions: perform navigation, clicks, text input, waits, and extraction based on the instruction.
• Output: return structured results as json, text, or markdown, optionally with screenshots.
• Failure handling: respond to timeouts, broken selectors, or anti-bot triggers by adjusting timeouts, reducing request rate, changing User-Agent, or locating elements via the accessibility tree.

Boundaries

It fits repetitive web interactions, data extraction, and page-testing tasks. It is not a good fit for complex business decisions, high-risk operations, or scenarios requiring strong authentication. Use it in line with target-site terms, request limits, and sensitive-data handling rules.