AgentsUse

Browser Automation

Browser Use

An open-source Python library that lets an LLM agent open pages, click, type, and fill forms from a plain task description.

Official siteDocumentationRepository

Quick decision

Best for
Builders who want an agent to handle varied web tasks from a task description, with custom tools and a choice of model.
Not ideal for
Repeatable, deterministic automation where the same steps must run identically every time. Direct browser code is easier to audit and cheaper to run.
Pricing model
The Python library is open source under MIT and free to use. Model inference and the optional hosted cloud browsers are paid separately. See the official docs for current hosted pricing.
Deployment
Self-hosted library. Hosted cloud (optional).
Authentication
Needs an LLM API key (for example an OpenAI, Anthropic, Google, or Browser Use key) in your environment. Cloud browsers use a Browser Use API key.
Review state
Source verified. Facts checked 2026-10-03. Not locally tested by AgentsUse.

What it does

Browser Use is an open-source browser agent for Python. You describe a task in plain language, connect a language model, and the agent opens a browser, reads the page, and carries out clicks, typing, and form fills.

The project offers three paths: a fully hosted cloud, a CLI that gives an existing coding agent browser access, and the Python library you run yourself with your choice of model and a local or cloud browser.

Custom tools can be registered as Python functions and passed to the agent, and the library can reuse a real Chrome profile for signed-in work. The hosted cloud adds managed browsers, stealth features, and recordings.

Verified capabilities

Browser control

  • Task-driven browsing Source verified

    An agent completes web tasks described in plain language, such as filling a form or extracting data to a file.

Data access

  • Form filling and data extraction Source verified

    Fill forms and pull structured data from pages as part of an agent run.

Agent integration

  • Custom tools Source verified

    Register Python functions as agent tools with the Tools class.

  • Choice of model Source verified

    Run with Browser Use models or bring provider keys for other models, including local models through Ollama per the docs.

  • CLI for coding agents Source verified

    A CLI path gives an existing agent browser access without writing library code.

Deployment

  • Local or cloud browsers Source verified

    Run against a local browser, a real Chrome profile, or hosted cloud browsers.

Quick start

Install the Python library

From the official README and PyPI page. Python 3.11 or newer is required, and you add your model key to a .env file.

uv add browser-use
# or: pip install browser-use

Environment variables (names only, use your own values)

  • OPENAI_API_KEY or another provider key
  • BROWSER_USE_API_KEY (only for Browser Use cloud and models)

Source: https://github.com/browser-use/browser-use. Examples use placeholders only. Never paste a real key into a profile, config file you share, or a ticket.

MCP support: Documented. The project docs cover MCP use, and the library can connect to local or cloud browsers. AgentsUse lists this as a documented path, not a locally tested one.

Works with

MCP

Documented MCP path in the project docs.

Packages

PyPI: browser-use. Installed with uv or pip.

Models

Bring your own provider key, use Browser Use models, or run a local model per the docs.

Deployment

Self-hosted library, CLI, or optional hosted cloud.

Only sourced support is listed. A missing framework means AgentsUse has not verified it yet, not that it cannot work.

Health and maintenance

GitHub stars
117,075 (checked 2026-10-03)
GitHub forks
12,923 (checked 2026-10-03)
Package
PyPI: browser-use, version 0.13.10
License
Open source, MIT
Maintainer
Browser Use

Stars and forks from the GitHub repository page; version from the PyPI release files (0.13.10, uploaded Sep 4, 2026). Maintenance signals only, not a quality rating.

Pricing and license

The Python library is open source under MIT and free to use. Model inference and the optional hosted cloud browsers are paid separately. See the official docs for current hosted pricing.

Official pricing or docs →

Limitations and safety

  • Results vary with the model and the site. Long tasks can drift, so keep a person reviewing important outcomes.
  • The library is free, but model calls and hosted browsers cost money. Budget per task before scaling up.
  • No independently checked benchmark is published by AgentsUse for this tool yet. The project publishes its own benchmark material, which AgentsUse has not reproduced.

Browser and data tools can read pages, fill forms, and download files. Start with a test account or read-only access, keep credentials in environment variables, and review agent actions before connecting anything that can spend money, send messages, or delete data.

Alternatives to Browser Use

Playwright

Choose Playwright when you want deterministic browser scripts and tests across three engines.

Tradeoff: You write and maintain the steps yourself. There is no model deciding the next action.

Stagehand

Choose Stagehand when you want an SDK with explicit act, observe, and extract calls in TypeScript, Python, or Go.

Tradeoff: You direct each step. It is less autonomous than a full task-driven agent.

Puppeteer

Choose Puppeteer when you want direct, low-level Chrome control from JavaScript.

Tradeoff: No agent loop or task planning. Every action is your code.

Related tools

Playwright

Browser Automation

Stagehand

Browser Automation

Puppeteer

Browser Automation

Common questions

What does Browser Use do for an AI agent?

An open-source Python library that lets an LLM agent open pages, click, type, and fill forms from a plain task description.

Is Browser Use open source?

Yes. This profile records the license as MIT from the official repository.

Does Browser Use support MCP?

Documented. The project docs cover MCP use, and the library can connect to local or cloud browsers. AgentsUse lists this as a documented path, not a locally tested one.

Sources and freshness

Last checked 2026-10-03. Verification label: source verified, which means public claims trace to the sources above. It does not mean AgentsUse ran the tool. Spotted an error? Send a correction. Back to the tools directory.