What it is

Browser Use is an open-source Python library and hosted platform that enables AI agents to control web browsers programmatically. It lives in the AI agent and browser automation ecosystem, built on Playwright and designed to let LLMs interact with websites as a human would—clicking, filling forms, solving CAPTCHAs, and navigating dynamic interfaces. The project provides three deployment paths: a fully hosted cloud service, a command-line interface for existing agents, and a local Python library for custom integrations.

The concrete problem it solves is the gap between high-level LLM reasoning and low-level browser control. Traditional automation tools require hardcoded scripts and struggle with unstructured, dynamic web content. Browser Use lets agents reason over tasks and adapt actions in real time, handling variability like changing UI layouts or interactive elements without manual intervention.

Key capabilities

• Execute multi-step browser tasks using natural language prompts via an LLM-powered agent loop
• Support for local Playwright browsers or cloud-hosted browsers via optional API
• Native integration with BU2, a model fine-tuned for browser automation, and OpenAI models
• Built-in CAPTCHA handling and stealth browser configurations to avoid detection
• Structured output support for extracting data from pages into JSON or Pydantic models
• CLI integration with existing agents (Claude Code, Codex, Hermes, OpenClaw, Cursor, Pi)
• Python library with async-first API for embedding agents in custom applications

Who uses it and how

• Developers embed the Python library into applications to automate tasks like appointment booking, price tracking, or form submission
• Teams use the hosted cloud API to scale agent operations with managed infrastructure, profiles, and data policies
• Researchers and hobbyists run the CLI to give existing AI agents browser access for interactive or scripted workflows

Getting started

Install via uv add browser-use (Python ≥3.11), set OPENAI_API_KEY or BROWSER_USE_API_KEY in .env, and run a Python script with Agent.run(). The cloud option requires no local browser setup; the CLI integrates with browser-use skill install in supported agents.

When to use it — and when not to

Use Browser Use when you need LLMs to handle unpredictable web interactions without brittle XPath or CSS selectors. It replaces parts of paid tools like Apify or Puppeteer-based services with open control and model flexibility. However, self-hosing requires managing Playwright browsers, and advanced stealth features (e.g., CAPTCHA bypass) may need cloud or BU2 access. Avoid if your tasks are purely static or API-based—this tool adds overhead where direct HTTP calls suffice.

Install or upgrade browser-use to the latest stable version with uv using Python 3.12, run `browser-use skill install` to register the skill, and connect it to my browser. If setup or connection fails, follow https://github.com/browser-use/browser-harness/blob/main/install.md.

uv add browser-use

# .env
OPENAI_API_KEY=your-key
# BROWSER_USE_API_KEY=your-key  # Optional: BU2 model or cloud browser

import asyncio

from browser_use import Agent, Browser, ChatBrowserUse, ChatOpenAI
from dotenv import load_dotenv

load_dotenv()

async def main():
    llm = ChatOpenAI(model='gpt-5.6-luna', reasoning_effort='xhigh')
    # llm = ChatBrowserUse(model='bu-2-0')  # Use BU2 instead; requires BROWSER_USE_API_KEY
    agent = Agent(
        task="Find the number of stars of the browser-use repo",
        llm=llm,
        # browser=Browser(use_cloud=True),  # Use a cloud browser; requires BROWSER_USE_API_KEY
    )
    history = await agent.run()
    print(history.final_result())

if __name__ == "__main__":
    asyncio.run(main())

uv run agent.py

from browser_use import Agent, ChatBrowserUse

llm = ChatBrowserUse(model='anthropic/claude-sonnet-4-6')  # or 'google/gemini-3-pro'
agent = Agent(task='...', llm=llm)