browser-use is a Python library and CLI that lets an AI agent operate a web browser to automate online tasks from natural language instructions. It provides code-level browser automation driven by an external LLM API service, which is required for operation.
Project overview
The project enables an AI agent to open pages, click, type, and fill forms based on described tasks, and ranks #1 on the Odysseys leaderboard with an 87.4% average.
Project type
AI Agent · AI Coding
Use cases
Automation
Deployment
Refer to project documentation
License
MIT
Best for
Developers using a Python library to automate web tasks at scale with any LLM.
Users needing an AI agent to execute browser actions like clicking, typing, and form filling from natural language instructions.
Key capabilities
Automates browser interactions by having an AI agent open pages, click buttons, type, and fill in forms based on described tasks.
A Python library allowing deep code-level integration to automate the web at scale with any LLM.
Allows users to add custom tools to extend the agent's capabilities via Python.
Extracts structured data from web pages, such as exporting followers as a CSV.
Limitations and risks
Chrome can consume a lot of memory and parallel agents are tricky to manage.
CAPTCHAs and browser fingerprinting can be an issue without stealth browsers or proxies.
An external LLM API service is required, meaning the data boundary is external and inference is not local-only.
Getting started
Setup is rated medium difficulty because it requires a Python environment and the provision of LLM API keys.
Install via uv add browser-use or pip install browser-use, add your LLM API key to the .env file, and define and run an Agent.
Alternatives and comparisons
Provides a thin, editable CDP browser harness where an LLM writes missing helper code during execution, contrasting with predefined agent actions.
Enables coding models to complete long-horizon web tasks by producing rerunnable Python scripts via a terminal, rather than relying on a one-action-at-a-time loop.
Converts websites and browser sessions into deterministic CLI interfaces through a logged-in Chrome session instead of AI-driven visual interaction.
GitHub project description: 🌐 Make websites accessible for AI agents. Automate tasks online with ease.
README: Browser Use lets an AI agent use a web browser the same way you do — it opens pages, clicks buttons, types, and fills in forms.
README: Browser Use is also **#1 on the [Odysseys leaderboard](https://odysseysbench.com/leaderboard)** with an 87.4% average
README: Want to automate the web at scale, from your own code, and with any LLM? Use the Python library:
README: **Use the CLI** if you already have an agent (Claude Code, Codex, Cursor, Hermes, OpenClaw, etc.) that you want to complete browser tasks for you.