jo-inc / camofox-browser

A stealth browser server with C++ fingerprint spoofing that gives AI agents reliable, token-efficient access to the real web.

活跃维护 MIT JavaScript Tracked
9.0k 964 5 天前
CIDockerNPM
ai-agent anti-bot antidetect-browser automation bot-detection browser-automation cloudflare-bypass headless-browser javascript nodejs playwright puppeteer scraping stealth-browser web-scraping

星标趋势

数据积累中,暂无足够数据生成趋势图

AI 分析

项目摘要

CamoFox Browser is a stealth headless browser server built for AI agents, leveraging Camoufox's C++ level Firefox fingerprint spoofing to bypass Cloudflare and bot detection. It provides a REST API with accessibility snapshots, stable element refs, and search macros as a drop-in Puppeteer/Playwright replacement optimized for agent workflows.

为什么值得关注

It addresses a critical pain point for AI agents — reliable web access — by patching Firefox at the C++ level rather than relying on JavaScript shims that become fingerprints themselves. The token-efficient accessibility snapshots and ultra-low idle memory footprint (~40MB) make it uniquely suited for agent infrastructure.

优势

  • C++ level fingerprint spoofing bypasses sophisticated bot detection that blocks Puppeteer/Playwright
  • Token-efficient accessibility snapshots reduce LLM context usage by ~90% vs raw HTML
  • Extremely lightweight (~40MB idle) enables deployment on Raspberry Pi or $5 VPS alongside agent stacks
  • Active development with 14 releases in 6 months and 31 contributors
  • Production-ready features: session isolation, proxy/geo-routing, structured JSON logging, cookie import

局限性

  • No examples directory despite being a drop-in replacement, slowing onboarding
  • Very new project (Jan 2026) with limited battle-testing in production environments
  • Firefox-only engine limits compatibility with sites requiring Chromium-specific behaviors

使用场景

  • AI agent web browsing and autonomous data extraction from protected sites
  • Large-scale web scraping with residential proxy rotation and anti-detection
  • Automated content aggregation pipelines feeding AI training or RAG systems
  • Authenticated session management for agents accessing user-specific web content
目标用户: AI agent developers and web scraping engineers who need reliable, undetectable browser automation that integrates cleanly with agent architectures
学习曲线:
分析模型:LongCat-2.0 | 分析时间:1 个月前