Technique Library
Explore AI attack techniques grouped by track, with lab density and challenge intensity.
Advanced Ctf
Capture-the-flag challenges testing advanced exploitation of AI systems — multi-step attacks, chained vulnerabilities, and creative problem-solving.
Agentic Ai Attacks
Attack autonomous AI agents — exploit tool-use chains, manipulate agent reasoning, and hijack multi-step workflows.
Ai Security
Core AI security fundamentals covering the full attack surface — from input manipulation to output exploitation across LLM applications.
Encoding Obfuscation
Bypass input filters using encoding tricks — Base64, Unicode, homoglyphs, and creative character substitution techniques.
Jailbreaking
Break out of AI safety guardrails — persuasion, role-play attacks, and techniques to bypass content policies and alignment.
Multi Agent Systems
Exploit multi-agent architectures — manipulate agent-to-agent communication, poison shared context, and attack orchestration layers.
Prompt Injection Basics
The foundation of LLM attacks — learn direct and indirect prompt injection, instruction override, and payload delivery fundamentals.
Rag Document Attacks
Poison retrieval-augmented generation systems — inject malicious documents, manipulate vector search, and exfiltrate data through RAG pipelines.
Supply Chain Attacks
Compromise AI supply chains — poisoned models, malicious packages, backdoored fine-tuning data, and dependency hijacking.
System Prompt Attacks
Extract and manipulate system prompts — leak hidden instructions, override developer constraints, and hijack AI application behavior.