# About OpenBrowse

> A browser agent you can inspect, operate and run on infrastructure you control.

Canonical URL: https://openbrowse.co/about
Markdown URL: https://openbrowse.co/about.md

## About OpenBrowse

OpenBrowse is an MIT-licensed, self-hosted alternative to Browser Use Cloud. It runs AI browser agents on your own Debian, Ubuntu or Raspberry Pi machine, while you choose the model provider, network path, operating budget and access controls. The agent drives a real browser and returns structured data through a documented v3 REST API. The first-party CLI is published on PyPI as openbrowse; the runtime source and releases live in the OpenBrowse GitHub repository.

Self-hosting is an operational trade-off, not a magic privacy label. You are responsible for securing the host, selecting providers and deciding who may use it. In return, browser profiles, session state and provider credentials stay with the instance you operate instead of being handed to a hosted automation platform. The live dashboard makes each run observable while it works, and the API exposes the status, output and failure information needed to operate it programmatically.

OpenBrowse at openbrowse.co is maintained by Lucas Johnston Kurilov. It is unrelated to openbrowse.org, openbrowse.ai and the npm package @openbrowse/mcp-server. Names that look alike do not indicate a shared service, maintainer, API or data boundary.

## What the project is for

### Ownership

Run the service on hardware and a network you select. There is no OpenBrowse per-task platform charge.

### Compatibility

Keep the v3 sessions and profiles workflow familiar to existing browser-use-sdk integrations.

### Evidence

Make agent work inspectable with structured output, a live view and explicit failure states rather than opaque completion claims.

## Read the details

Start with the installation guide, review how to expose an instance safely, or inspect the source code on GitHub.