# camofox-browser

An anti-detection headless browser server for AI agents, built on the Camoufox Firefox fork, exposed as a REST API that returns accessibility snapshots and stable element references instead of raw HTML.

camofox-browser is an anti-detection browser server for AI agents, built on top of Camoufox, a Firefox fork that spoofs fingerprinting signals such as navigator.hardwareConcurrency, WebGL renderers, AudioContext and WebRTC at the C++ implementation level before JavaScript can see them. It wraps that engine in a REST API designed for agents rather than for humans.

Instead of returning raw HTML, it serves accessibility snapshots with stable element references for clicking, plus search macros for common sites such as Google, YouTube, Amazon and Reddit. It supports session isolation with separate cookies and storage per user, import of Netscape-format cookie files for authenticated browsing, file uploads from a configured directory, residential proxy routing with matching locale and timezone, structured JSON logging, YouTube transcript extraction, optional screenshots alongside a snapshot, and paginated handling of large pages.

It is designed to run alongside other services on constrained hardware, with a lazy browser launch and idle shutdown that keep memory around 40MB when idle. It is written in JavaScript and licensed under MIT.

Website: https://github.com/jo-inc/camofox-browser