antibrow/anti-detect-browser-skills

Launch and manage anti-detect browsers with unique real-device fingerprints for multi-account operations, web scraping, ad verification, and AI agent automation.

What it solves

This project provides a set of "skills" for AI agents (specifically Claude Code) to control a browser that mimics real devices to avoid bot detection. It solves the problem of headless browsers being easily flagged by anti-bot vendors because they lack consistent, real-world hardware and network fingerprints.

How it works

Instead of using JavaScript patches to hide its identity, the project uses a modified Chromium kernel where fingerprints (Canvas, WebGL, WebGPU, audio, fonts, etc.) are handled at the C++/Blink level. This ensures that Web Workers and the main thread return identical, consistent data. Each browser profile is assigned a unique, coherent persona sampled from a real machine, and network settings like timezone and WebRTC identity are automatically synced with the proxy's exit IP.

Who it’s for

  • QA and Bot-Detection Testers: Those testing their own site's anti-fraud stacks.
  • Data Scrapers: Users needing to collect public data from sites that require sessions or use heavy bot detection (e.g., Amazon, Google, X).
  • AI Agent Developers: Developers wanting to give an AI agent a persistent, logged-in browser identity via MCP (Model Context Protocol).
  • Automation Engineers: Those managing multiple authorized accounts across different regions for pricing or ad verification.

Highlights

  • Kernel-Level Fingerprinting: Avoids detection by implementing identity at the browser engine level rather than via script injection.
  • MCP Agent Integration: Allows AI agents to launch and control the browser directly via tool calls without writing code.
  • ** uma-Account Scraping**: Provides per-site JSON commands (recipes) that can be run across multiple profiles simultaneously (fanout).
  • Dual SDKs: Full support for both JavaScript/TypeScript and Python.
  • Cross-Platform Support: Works on Windows, macOS (Intel/Silicon), and Linux (x64/arm64), including Docker.
  • Isolation Verification: Includes a specific skill for verifying that profiles are truly isolated through concrete consistency checks.

Related

  • Project
    Winson-030/dify-kubernetes

    A collection of Kubernetes deployment manifests that enable users to easily deploy and deploy Dify on a Kubernetes cluster.

  • Project
    Nebulock-Inc/agentic-threat-hunting-framework

    A Python CLI framework that stores threat‑hunting investigations as markdown using the LOCK pattern, makes them searchable, and optionally lets LLM agents generate research, run queries, and automate parts of the hunt.

  • Project
    evidentloop/sopify

    Sopify is a development process protocol for AI coding that prevents premature coding by enforcing a plan-first workflow and storing resumable project plans in git-tracked files.

  • Project
    gethasp/hasp

    A local secret broker for coding agents that keeps sensitive credentials in an encrypted vault and prevents them from entering the agent's context.