amazon-science/Cyber-Zero

Cyber-Zero: Training Cybersecurity Agents Without Runtime

50
/ 100
Established

This project helps cybersecurity professionals develop and train autonomous agents to solve Capture The Flag (CTF) challenges. It takes publicly available CTF write-ups and, using simulated interactions, generates realistic training scenarios without needing actual execution environments. The output is high-quality training data for LLM-based cybersecurity agents, allowing practitioners to build more capable automated systems for vulnerability finding.

Use this if you are a cybersecurity researcher or red teamer looking to train advanced AI agents to identify vulnerabilities and solve CTF challenges efficiently without the overhead of real-world environment setup.

Not ideal if you need to test agents against live, real-time network environments or if your primary focus is manual penetration testing.

cybersecurity-training CTF-challenges vulnerability-analysis red-teaming security-automation
No Package No Dependents
Maintenance 10 / 25
Adoption 9 / 25
Maturity 15 / 25
Community 16 / 25

How are scores calculated?

Stars

77

Forks

12

Language

Python

License

Last pushed

Feb 13, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/amazon-science/Cyber-Zero"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.