The General-Purpose Agentic Browser
Give your AI agent a real browser. 3.2x more token-efficient than Playwright MCP. 6x more than Chrome DevTools MCP.
Built different from the ground up.
Get Started in 3 Steps
From zero to autonomous browsing in minutes.
1
Install
pip install openbrowser-ai
2
Write a Task
from openbrowser import CodeAgent, ChatGoogle
from dotenv import load_dotenv
import asyncio
load_dotenv()
async def main():
agent = CodeAgent(
task="Find the #1 post on Show HN",
llm=ChatGoogle(),
)
await agent.run()
asyncio.run(main())
3
Watch It Browse
The agent navigates, clicks, types, and extracts data autonomously. Watch it work in real-time through the live browser view.
See OpenBrowser in Action
Watch a real agent browse the web autonomously.
Works With Your Favorite Tools
OpenBrowser integrates with any MCP-compatible client.
Get Early Access
Be the first to try the hosted version of OpenBrowser.