Browser Tool

About

Browser for AI Agents.

The vision for Browser use is to provide a browser for AI Agents to interact with the web while also allowing for the user to interact with the browser and have a live feed of agent actions similar to OpenAI's Operator.

Features

• Initiate a browser session
• Interact with the browser
• Have a live feed sent through websockets

Demo

Browser Tool Demo

Running Locally

Prerequisites

• Python 3.8 or higher
• Node.js 16.x or higher
• pip
• npm

Backend Setup

Navigate to backend directory

cd backend

Create and activate virtual environment

python -m venv .venv
source .venv/bin/activate

On Windows use:

venv\Scripts\activate

Install dependencies

pip install -r requirements.txt

Start the backend server

python app/main.py

Frontend Setup

Navigate to frontend directory

cd frontend

Install dependencies

npm install

Start the development server

npm run dev

Vision

A Browser where AI agent can partner with the user to browse the web and perform tasks.

Contributing

We love contributions! Feel free to open issues for bugs or feature requests. To contribute to the docs, check out the /docs (not created yet TODO: thoufic) folder.

Citation

@software{browser_tool,
author = {[Thoufic]},
title = {Browser Tool},
year = {2025},
publisher = {GitHub},
url = {https://github.com/browser-use/browser-use}
}