browser-use
Installation
SKILL.md
Browser Use — AI Browser Automation Agent
Overview
Browser Use gives an LLM a browser: the agent reads the page (DOM plus screenshots), then clicks, types, scrolls, switches tabs and extracts data until the task is done. Checked against browser-use 0.13.10 (4 September 2026). Three ways to use it:
- Python library (this skill's focus):
Agent(task=..., llm=...)running locally, with your choice of model. - CLI (
browser-use, built on Browser Harness): lets a coding agent such as Claude Code run Python helpers against your own Chrome with its logins. - Browser Use Cloud: hosted agents and stealth browsers, paid per use.
It no longer sits on LangChain. Models are Browser Use's own wrappers (ChatOpenAI, ChatAnthropic, ChatGoogle, ChatOllama, ChatBrowserUse, and more), and the old BrowserConfig and output_model= names are gone.