VulnBot: Autonomous Penetration Testing for a Multi-Agent Collaborative Framework

Table of Contents

• Overview
• Quick Start
• Citation
• Contact

---

Overview

Note:

• ⭐ If you find this project useful, please consider giving it a STAR! ⭐
• If you encounter any errors or issues, feel free to open an issue or submit a pull request.

VulnBot is an advanced automated penetration testing framework that utilizes Large Language Models (LLMs) to replicate the workflow of human penetration testing teams within a multi-agent system. This innovative approach enhances efficiency, collaboration, and adaptability in security assessments.

This RAG implementation is based on Langchain-Chatchat. Special thanks to the authors.

---

Quick Start

Prerequisites

Ensure your environment meets the following requirements before proceeding:

• Programming Language: Python 3.11.11
• Package Manager: Pip

Installation

Install VulnBot using one of the following methods:

Build from Source

1. Clone the VulnBot repository:

   git clone https://github.com/KHenryAegis/VulnBot

2. Navigate to the project directory:

   cd VulnBot

3. Install the dependencies:

   pip install -r requirements.txt

Configuration Guide

Before initializing VulnBot, you need to configure system settings. Refer to the Configuration Guide for detailed instructions on modifying:

• Kali Linux configuration (hostname, port, username, password)
• MySQL database settings (host, port, user, password, database)
• LLM settings (base_url, llm_model_name, api_key)
• Enabling RAG (set enable_rag to true and configure milvus and kb_name)

Initialize the Project

Before using VulnBot, initialize the project:

python cli.py init

Start the RAG Module

python cli.py start -a

Run VulnBot

To execute VulnBot, use:

python cli.py vulnbot -m {max_interactions}

Replace {max_interactions} with the desired number of interactions.

---

Citation

If you use VulnBot for academic purposes, please cite our paper:

@misc{kong2025vulnbotautonomouspenetrationtesting,
      title={VulnBot: Autonomous Penetration Testing for a Multi-Agent Collaborative Framework}, 
      author={He Kong and Die Hu and Jingguo Ge and Liangxiong Li and Tong Li and Bingzhen Wu},
      year={2025},
      eprint={2501.13411},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2501.13411}, 
}

---

Contact

If you have any questions or suggestions, please open an issue on GitHub. Contributions, discussions, and improvements are always welcome!