This project provides an AI-driven platform for automating CTF (Capture The Flag) challenges, designed for security researchers and participants. It leverages Large Language Models (LLMs) and a multi-agent system to automate complex vulnerability discovery and exploitation, significantly enhancing efficiency in security competitions.

How It Works

The core architecture employs a multi-agent system comprising five specialized agents: explorer, scanner, solutioner, executor, and actioner. These agents collaborate, with LLMs driving decision-making, vulnerability analysis, and exploit code generation. This approach enables end-to-end automation, from intelligent page exploration and multi-dimensional vulnerability scanning (including XSS, SQLi, command injection, LFI, IDOR) to cross-scenario exploitation and automated exploit execution. The system features a Flask backend for API services and a separate frontend for task management.

Quick Start & Requirements

• Environment: Python 3.8+, pip, SQLite3.
• Installation: Clone the repository, navigate to the agent directory, and run pip install -r requirements.txt.
• Configuration: Edit agent/config/config.py to set LLM API keys (e.g., DeepSeek, Tencent) and the backend server URL.
• Execution: Run python flaghunter.py from the agent directory.
• Docker Deployment: Navigate to the server directory and run docker-compose up -d. Access the frontend at http://localhost:85 and the backend API at http://localhost:5000.

Highlighted Details

• Multi-Agent Framework: Five distinct agents (explorer, scanner, solutioner, executor, actioner) manage specialized tasks for robust automation.
• LLM Integration: Utilizes LLMs for intelligent analysis, exploit generation, and strategic decision-making throughout the CTF challenge.
• Comprehensive Vulnerability Support: Automates discovery and exploitation for common web vulnerabilities like XSS, SQL injection, command injection, LFI, and IDOR.
• Extensible Tooling: Features built-in tools (HTTP requests, Python/Shell execution) and supports custom extensions for decoding and payload generation.

Maintenance & Community

The project originated from the Tencent Cloud AI Penetration Hackathon, developed by the xjtuHunter team, securing second place. The core maintainer is 九暑 (passerW). No specific community channels (like Discord or Slack) or detailed roadmap information are provided in the README.

Licensing & Compatibility

The project is released under the MIT License, which permits broad use, including commercial applications and linking within closed-source projects.

Limitations & Caveats

This tool is strictly intended for legal CTF competitions and authorized security research. Use on unauthorized systems is prohibited. The functionality relies on obtaining and configuring API keys for specific LLM providers.