I wrapped browser-use in a REST API with features I needed for production use.

browser-use is great but it's a Python library. I wanted an API I could call from anywhere with proper session management, visibility into what's happening, and extensibility.

Features: - VNC streaming: watch the browser live at /vnc.html - Session management: launch browsers, reuse across tasks, save profiles - Custom tools: register HTTP endpoints the agent can call (APIs, webhooks, etc.) - Task control: start/stop/pause, step-by-step execution updates - 15+ LLMs: GPT, Claude, Gemini, Groq – swap with one parameter

Dev branch also has R2 backup for browser profiles (useful for serverless).

    docker run -d -p 8080:8080 reqeique/browser-use-api:dev

Couldn't find an existing API layer for browser-use with these features, so I built one. MIT licensed.

Happy to hear what's missing or what you'd want from something like this. MCP support is coming soon.