Kim-Hammar/awesome-rl-for-cybersecurity

What it solves

This project addresses the challenge of finding high-quality, specialized environments and resources for training and evaluating reinforcement learning (RL) agents specifically for cybersecurity tasks. It filters out general machine learning resources to focus exclusively on RL applications.

How it works

It serves as a curated directory of simulation environments, research papers, books, blog posts, and talks. The environments listed provide standardized interfaces (often compatible with OpenAI Gym/Gymnasium) that allow researchers to simulate network attacks, defensive strategies, and malware interactions, enabling agents to learn through trial and error in controlled, safe settings.

Who it’s for

This list is designed for cybersecurity researchers, AI practitioners, and students who are developing autonomous agents for tasks like network defense, automated penetration testing, and traffic classification.

Highlights

• A comprehensive collection of cyber-security-specific RL environments, including CSLE, CybORG++, and Microsoft CyberBattleSim.
• Categorized resources ranging from academic papers and theses to practical blog posts and technical documentation.
• Focuses exclusively on reinforcement learning, excluding general machine learning methods.
• Includes tools for both offensive (penetration testing) and defensive (intrusion response) AI research.