browser-use/browser-use's Claude Code Plugin for Browser Automation

What is browser-use/browser-use, actually

browser-use/browser-use is the repository behind Browser Use, a project built so an AI agent can use a web browser the way a person does: opening a page, clicking a button, typing into a field, filling out a form, from a plain description of the task rather than a script. The repository carries 111,539 GitHub stars, and Flowy lists it under the automation category. The Claude Code plugin is a small, specific slice of that repository: 6 skills, every one of them a skill and nothing else. Flowy indexes what the repository ships. It does not own, host, or run the browser automation itself.

What is inside

All 6 components are skills, and they split into three different jobs. browser-use drives a real browser directly, and qa and remote-browser also work against a live browser, each for a narrower purpose. open-source and cloud are not automation at all, they are documentation: reference material for writing code against the project's two surfaces, the Python library and the hosted API. x402 is neither of those: it is a payment setup skill for Browser Use's Cloud service.

Which skills actually drive a browser

browser-use is the core of the plugin: direct browser control over CDP, Chrome's DevTools Protocol, for automation, scraping, testing, screenshots, and general site or app work. This is the skill to invoke when you want Claude Code to open a page and act on it, rather than read about how.

qa narrows that to grading instead of driving. It tests a website or web app and returns a 1 to 5 quality score, 5 for flawless and 1 for broken, backed by evidence instead of a bare number. Its own description calls out local development by name, giving testing a local server such as localhost:5173 as a worked example, so a page that is not deployed anywhere yet still qualifies.

remote-browser solves a different problem: an agent with no display of its own. It controls a local browser from a sandboxed remote machine, for when the agent is running somewhere headless but still needs to navigate a site, fill in a form, take a screenshot, or expose a local dev server to the outside world through a tunnel.

Which skills are pure documentation

Two of the six skills exist only as reference material, split by where your code actually runs. open-source documents writing Python against the browser-use library itself: its Agent, Browser, and Tools building blocks, code that imports from browser_use, sandbox deployment, and which language models the library supports.

cloud documents the alternative: Browser Use Cloud, the hosted version of the same idea. It covers the Cloud REST API in both its v2 and v3 forms, the SDK available for Python or TypeScript, authenticating with an API key header, and working with cloud sessions. Reach for open-source while you are running the browser yourself in Python, and for cloud when you would rather call a hosted service and let Browser Use run the browser for you.

What does x402 do

x402 is the narrowest skill in the plugin: it sets up payment for Browser Use Cloud through the x402 protocol, meaning you pay per request from a crypto wallet holding USDC on the Base mainnet, with no signup and no API key. Its description frames one setup path as "just use it," which configures a wallet so that you, or Claude Code acting on your behalf, can run cloud browser tasks paid directly from it. This only matters if wallet based, pay as you go access is what you actually want instead of a signed up Cloud account.

Who this is actually useful for

The 6 skills split cleanly by role. Someone scraping or automating a site locally reaches for browser-use. Someone grading a build reaches for qa. An agent stuck in a sandbox without a display reaches for remote-browser. A developer writing Python against the library reaches for open-source, and one calling the hosted API instead reaches for cloud. Anyone funding Cloud usage from a crypto wallet rather than a signup flow reaches for x402.

How do I install it

Install from the listing page linked at the top of this guide. Flowy indexes browser-use/browser-use from its public repository. It does not own, host, or modify the project, and all 6 skills above stay on their author's own releases in the source repository.