// about the maze
Security, learned by walking the paths.
The Labyrinth is a collection of hands-on security labs: real attacks, real detections, real write-ups. Each lab is a maze to work through, from first probe to final report.
Topics span LLM and AI security, network security, SASE and zero trust, and whatever comes next.
$ labyrinth ls ai-gateway LLM security ● live netsec network sec. ○ soon sase zero trust ○ soon detect detection eng. ○ soon $ labyrinth enter ai-gateway ! sign-in required → redirecting to lab.thelabyrinth.app
// learn the walls
Start here. No sign-in needed.
Two quick ways into LLM security: attack a live simulation, or work through the OWASP risk list. Useful whether you are new to the field or reviewing.
Spot the attack interactive demo
Run prompt injection, system prompt extraction, and data-leak attacks against a mock chatbot. See how an AI gateway detects each one, and why it works.
- Prompt injection
- Prompt leakage
- PII / DLP
- Indirect injection
The LLM Top 10, explained OWASP 2025 list
The OWASP 2025 list of the most critical LLM application risks. A one-line summary per risk, with mechanism, example, and controls on demand.
- 10 risks
- Example attacks
- Controls
- Lab coverage
// choose a path
Labs
AI Gateway LLM Security Lab
OWASP LLM Top 10Attack an LLM-protected app, catch it from the blue-team side, then write it up.
- Prompt injection attack waves
- Blue team detection
- Incident report write-up
Network Security segmentation & firewalling
A new path is being carved. Check back soon.
SASE & Zero Trust secure access
A new path is being carved. Check back soon.
Detection Engineering logs to alerts
A new path is being carved. Check back soon.