browser-use

Automates browser control via Chrome DevTools Protocol for scraping, testing, and screenshots.

Updated Jul 31, 2026
One-click install
npx skills add https://github.com/AarnavBaddam/skills --skill browser-use-aarnavbaddam
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: browser-use
Source: https://github.com/AarnavBaddam/skills/tree/main/browser-use
Command: npx skills add https://github.com/AarnavBaddam/skills --skill browser-use-aarnavbaddam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Manually driving a browser for automation, scraping, or testing is slow and error-prone. This Skill gives an AI agent direct control of a local or cloud-hosted Chrome browser through the Chrome DevTools Protocol (CDP), enabling navigation, clicking, screenshots, and DOM extraction without writing low-level automation code. ## Core Features & Use Cases - Direct CDP Browser Control: Navigate pages, click by coordinates, run JavaScript, capture screenshots, and inspect the DOM through pre-imported helper functions. - Local and Cloud Browsers: Attach to a running local Chrome instance or spin up isolated cloud browsers for parallel tasks and captcha-sensitive scraping. - Interaction Skill References: Consult dedicated guides for tricky mechanics like iframes, shadow DOM, downloads, drag-and-drop, and network requests. - Use Case: Imagine you need to scrape pricing data from a bot-protected site. Start a cloud browser, navigate to the page, screenshot the state, extract the data with JavaScript, and stop the remote browser when finished. ## Quick Start Ask the agent to open a website in the browser, take a screenshot, and extract the page's main content using the browser-use command.

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 Chrome DevTools Protocol?

Run the browser-use command with a heredoc containing helper calls like new_tab(url), click_at_xy(x, y), and js(...) for DOM work. The harness connects to Chrome's CDP endpoint automatically via ensure_daemon() before executing your code.

How do I take screenshots and click elements in an automated browser?

Call capture_screenshot() first to see the visible state, read the pixel coordinates of the target, then call click_at_xy(x, y) and screenshot again to confirm. After any navigation, call wait_for_load() before interacting.

Can I run parallel browser automation tasks in the cloud?

Yes, Browser Use cloud browsers are isolated managed Chrome instances, one per task, avoiding tab and focus conflicts in a shared local browser. Authenticate with browser-use auth login, then start a named remote daemon and target it with the BU_NAME environment variable.

Why can't the browser automation connect to my local Chrome?

Local control requires Chrome remote debugging to be enabled. Run browser-use --doctor for diagnostics, then open chrome://inspect/#remote-debugging, tick Allow remote debugging for this browser instance, and approve any Chrome permission popup.

When should I use a cloud browser instead of local Chrome?

Use cloud browsers for parallel concurrent tasks or when captchas and bot-blocking are likely, since they run with clean managed IPs and stealth settings. Remember to stop remote daemons when done because they bill until stopped or timed out.