A Chromium fork that puts AI agents in the driver’s seat

What it does

BrowserOS is a Chromium fork with a built-in agent platform. It gives you 53+ browser automation tools—clicking, typing, navigating, extracting data—driven by natural language instructions. You can bring your own API keys for Claude, OpenAI, Gemini, and others, or run local models through Ollama and LM Studio so your browsing data stays on your machine.

The interesting bit

Instead of duct-taping a chatbot onto a toolbar, BrowserOS patches Chromium itself and runs a Bun-based MCP server alongside it. That means Claude Code—or any MCP client—can remote-control your browser, while persistent memory, scheduled tasks, and a visual workflow builder all live in the same binary.

Key highlights

• Native AI agent with 53+ browser automation tools and persistent memory across sessions
• MCP server support, so external tools like Claude Code can drive the browser
• Visual workflow builder and scheduled tasks for repeatable automations
• Local-first option via Ollama and LM Studio; also supports OAuth for ChatGPT Pro, GitHub Copilot, and API keys for major cloud providers
• Bundled ad blocking via uBlock Origin with Manifest V2 support, plus vertical tabs and 40+ app integrations through MCP

Caveats

• Building the browser from source requires roughly 100 GB of disk space and patches Chromium directly, so contributions to the browser core are heavy lifting compared to the TypeScript/Go agent layer.
• The project is AGPL-3.0 licensed, which may affect how you redistribute modified versions.
• Download badges label the macOS, Windows, and Linux builds as beta.

Verdict

Worth a look if you want an agentic browser that keeps data local and plays nice with MCP tooling. Skip it if you are perfectly happy with separate automation scripts and a standard Chromium build.