browser-use-expert_skill

Overview

This skill describes a Senior Web Automation Engineer who specializes in the browser-use CLI, Python library, and agentic web orchestration (v0.11.4+). It focuses on reliable, persistent browser sessions, multi-step workflows, and production-ready automation patterns for January 2026 standards. Primary model recommendation is ChatBrowserUse via BROWSER_USE_API_KEY for best accuracy and performance.

How this skill works

The skill prefers the browser-use CLI for direct, persistent interactions and the Python library for conditional logic and loops. It inspects page state with browser-use state to get element indices, drives interactions with click/input/select/keys commands, and verifies actions with screenshots or repeated state checks. Use remote cloud browsers for stealth and reliability and always close sessions to avoid resource leakage.

When to use it

• Search the web, extract structured data, or scrape dynamic pages requiring interaction
• Automate multi-step flows like login → navigate → fill checkout → confirm
• Monitor a site for changes, new content, or price updates with persistent sessions
• Debug interactive workflows locally using headed mode to observe failures
• Run production tasks that need anti-detection and captcha resistance via remote browsers

Best practices

• Default to ChatBrowserUse (BROWSER_USE_API_KEY); fall back to ChatGoogle or ChatAnthropic only if unavailable
• Prefer the browser-use CLI for atomic, fast steps; use browser-use python for complex logic
• Always run browser-use state before interacting to get accurate element indices
• Use --browser remote for production or captcha-prone tasks and --headed when debugging locally
• Always close sessions with browser-use close or browser-use close --all to prevent leaks
• Install via uv (uv pip install browser-use) and set PLAYWRIGHT_BROWSERS_PATH when needed

Example use cases

• Autonomously fill a checkout form across multiple pages, verify receipts, and close the session
• Extract and normalize price data from JavaScript-heavy sites while preserving login state
• Continuously monitor a news or product page and trigger downstream alerts on changes
• Use browser-use python to run conditional logic: loop through results, retry on failures, and store outputs
• Run headless remote browsers for scheduled scraping to avoid local infrastructure and captchas