camofox-browser
SHA-256Anti-detection browser server for AI agents, powered by Camoufox (Firefox fork). It bypasses Cloudflare and bot detection via C++-level fingerprint spoofing. A drop-in Puppeteer/Playwright replacement that avoids stealth-plugin tells.
Stealth headless browser for AI agents with C++-level fingerprint spoofing, bypasses Cloudflare and bot detection.
Core Features
- C++-level anti-detection bypasses Cloudflare and most bot mitigation
- Accessibility snapshots ~90% smaller than HTML — token efficient
- Stable element refs (e1/e2/e3) for reliable interaction
- Runs on a Raspberry Pi or $5 VPS with ~40MB idle memory
- Session isolation and cookie import for authenticated browsing
What It Can't Do
- •First run downloads a ~300MB Camoufox binary — be patient. Do NOT run docker build directly; always use make up or build.ps1 to fetch binaries first, or the build will fail. Playwright-specific env vars like PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD can silently break the Camoufox download — see README for workarounds (--ignore-scripts or CAMOUFOX_EXECUTABLE).
Use Cases
- AI agents needing undetected access to login-walled or scraped sites
- Developers replacing Puppeteer/Playwright for token-efficient web automation
Detailed Introduction
camofox-browser is a stealth headless browser server built for AI agents and web automation. Unlike Playwright or Puppeteer with stealth plugins—which are easily fingerprinted—Camoufox patches Firefox at the C++ level, spoofing navigator.hardwareConcurrency, WebGL renderers, AudioContext, screen geometry, and WebRTC before JavaScript can see them. This makes it effective against Google, Cloudflare, and most bot detection systems. The project wraps this engine in a REST API designed for agent workflows: token-efficient accessibility snapshots (about 90% smaller than raw HTML), stable element references (e1, e2, e3) for reliable clicks and typing, session isolation, cookie import, proxy with GeoIP support, YouTube transcript extraction, and search macros for common sites. It runs on a Raspberry Pi or a $5 VPS with around 40MB idle memory, features Docker deployment, interactive VNC login, and structured extract endpoints with JSON Schema. A drop-in replacement that simpler and more secure than stealth plugins.
Tags
Getting Started
Download installer
Click the button above to download the installer for your system
Install the software
Double-click the downloaded installer and follow the prompts
Clone the repo: git clone https://github.com/jo-inc/camofox-browser
Install and start: npm install && npm start (first run downloads Camoufox ~300MB)
Open http://localhost:9377 to use the API or interactive docs
- Clone the repo: git clone https://github.com/jo-inc/camofox-browser
- Install and start: npm install && npm start (first run downloads Camoufox ~300MB)
- Open http://localhost:9377 to use the API or interactive docs
SHA-256 checksum verified
Checksum extracted from GitHub official Release page
SHA256 Checksum
5720d45b894ce1770543de024c6f10d514b38be560fa2dc3226b3d8586caf672This checksum is extracted from the GitHub Release page. Verify file integrity after download.
All SHA-256 checksums on this platform are extracted from the project's official GitHub Release page, without any modification. You can independently verify them on the GitHub Releases page.
Open Source Transparency
View GitHub SourceUninstall Info
Delete the project folder to remove the server. For any downloaded Camoufox binaries and caches, delete ~/.cache/camoufox and related ms-playwright folders.
No Extra Dependencies
Ready to use after download. No additional runtime required.
Similar Projects
ollama
Ollama lets you download, run, and manage large language models locally. One command, multiple platforms, endless possibilities.
ComfyUI
ComfyUI is a modular AI engine with a node-based graph interface for generating images, video, 3D models, and audio. Unlike traditional webui tools, it gives pixel-level control over every model and parameter, making it the preferred choice for professionals.
llama.cpp
High-performance LLM inference engine in C/C++ with minimal dependencies, supporting quantized models (1.5–8 bit) and diverse hardware (Apple Silicon, CUDA, Vulkan, etc.).