browser-use/browser-use
A tool that lets AI click buttons and type into websites just like a human, to finish tasks for you
browser-use is a Python library that lets an AI agent open a real web browser and click, type, and navigate pages on its own. You describe a task in plain language, and the agent operates the browser to complete things like filling out job applications or extracting follower data. It's free and open source, with a paid cloud version available for heavier or more reliable use.
What it does
- It can be installed as a 'skill' inside existing AI coding agents like Claude Code or Cursor, giving them the ability to control a browser.
- In Python code, you connect an Agent to an LLM (large language model) to run tasks at scale, on a schedule, or in parallel.
- On the project's own 100-task real-world benchmark and the Odysseys leaderboard covering 200 long-horizon web tasks, it reports an 87.4% average, ahead of computer-use agents from OpenAI, Anthropic, Google, and Microsoft.
- The open-source version runs free on your own machine with full control over which LLM to use, while the cloud version adds stealth browsing, CAPTCHA solving, and large-scale parallel execution.
- It supports custom tools and reusing existing Chrome login profiles, making it practical for real automation work involving authentication.
Why it matters
Being able to hand off repetitive browser work—form filling, data scraping, tasks requiring login—to an AI agent means developers can build automation, QA, or scraping pipelines with much less manual coding. It's MIT-licensed and free to start with, and can scale up to a paid cloud service without changing much code, which matters for anyone moving from a prototype to production.
Terms in this repo
- Agent · An AI program that takes a goal and decides its own sequence of actions to reach it
- LLM · Large language model, the AI trained on huge amounts of text that powers the agent's decisions
- MCP · A standard protocol for connecting AI tools to external services
- CAPTCHA · A test used to tell humans apart from bots on websites
- benchmark · A standardized test used to compare the performance of different systems
Repository description (English)
🌐 Make websites accessible for AI agents. Automate tasks online with ease.
Open on GitHubTrending repos
- tt-a1i/archifyAn AI agent skill that turns a described codebase into a verified, shareable architecture diagram inside chat
- bookorbit/bookorbitA self-hosted book server that syncs your reading progress across web, Kobo, and KOReader
- ConardLi/garden-skillsA skill pack that hands AI coding agents ready-made expertise for design, video, images, and search
- K-Dense-AI/scientific-agent-skills163 how-to manuals that turn AI coding assistants into science research helpers
- DietrichGebert/ponytailA ruleset that teaches AI coding agents to say no to writing code
- liustack/modlensA plugin that lets text-only AI models like DeepSeek 'see' pasted images by turning them into structured text data
- marin-community/marinMarin is an open-source platform that publishes the entire process of building large language models, not just the final model
- TauricResearch/TradingAgentsA framework where multiple LLM agents split into analyst, researcher, trader and risk-management roles, debate, and decide stock trades
Latest from METAL LAB
- Ox Alpha turns out to be GLM-5.3-Flash
- Martin Casado says AI has erased startups' distribution barrier
- OpenAI introduces Runme, which uses Codex to document repetitive work
- Stable Diffusion Maker Raises $76M From Music, Game Studios
- Meta's paid agent Hatch to launch alongside new Watermelon model in October