browser-use

Automate browser interactions, testing, and page capture through the playwright-cli command line.

Updated Jul 10, 2026
One-click install
npx skills add https://github.com/Ilm-Alan/browser-use --skill browser-use-ilm-alan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: browser-use
Source: https://github.com/Ilm-Alan/browser-use
Command: npx skills add https://github.com/Ilm-Alan/browser-use --skill browser-use-ilm-alan

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @playwright/cli, and includes references (resource) components.

What problem does it solve? Automating real browser behavior—clicking, filling forms, logging in, scraping rendered pages, and debugging UI bugs—normally requires writing and maintaining Playwright scripts by hand. This Skill lets an agent drive a live browser directly through the playwright-cli binary, including attaching to the user's signed-in Chrome for authenticated tasks. ## Core Features & Use Cases - Full browser control: Navigate, click, fill forms, run JavaScript in-page, take ARIA snapshots, screenshots, and PDFs, manage tabs, cookies, and localStorage. - Authenticated sessions: Attach to the user's real Chrome via the Playwright MCP Bridge extension or CDP, bypassing Google/OAuth blocks on fresh automated browsers. - Network and debugging tools: Inspect and mock network requests, record traces and video, read console output, and run or debug Playwright tests with --debug=cli. - Use Case: Ask the agent to log into your YouTube Studio dashboard, update a video's tags, and screenshot the result—it attaches to your real Chrome, drives the shadow-DOM UI with native keystrokes, and captures proof of the change. ## Quick Start Ask the agent to open a website in the browser, fill in the login form, and take a screenshot of the resulting page.

Frequently Asked Questions about browser-use

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate browser actions with playwright-cli?

Run playwright-cli commands like open, click, fill, and snapshot to drive a browser session directly from the shell. Each command returns an ARIA snapshot with element refs (e.g. e5) that you pass to subsequent interaction commands.

How do I automate tasks in my logged-in Chrome session?

Use playwright-cli attach --extension=chrome with the Playwright MCP Bridge extension installed and PLAYWRIGHT_MCP_EXTENSION_TOKEN set. This drives your real Chrome profile with existing cookies and logins, avoiding Google/OAuth blocks on fresh automated browsers.

Can playwright-cli mock network requests during automation?

Yes, the route command intercepts URL patterns and returns custom status codes, bodies, or headers, and unroute removes them. For conditional logic or response modification, use run-code with page.route in a Playwright function.

Why does querySelectorAll fail on shadow DOM pages like YouTube Studio?

querySelectorAll does not pierce shadow roots, so elements inside web components return null. Use the ARIA snapshot command instead, which pierces shadow DOM and provides stable refs, and prefer native clicks and keystrokes over synthetic events.

How do I debug a failing Playwright test interactively?

Run the test with npx playwright test --debug=cli in the background, then attach with playwright-cli attach using the printed session name. The paused page can be explored with snapshot, console, and requests, and every action generates TypeScript you can paste into the test.