$ whoami

Hack AI Agents

Learn to audit, exploit, and secure AI agent systems. From prompt injection to supply chain attacks — master the full attack surface of modern AI agents with hands-on labs.

Start Hacking →

$ nmap -sV target-agent.api

Scanning attack surface...

[!] Found: Prompt injection vector

[!] Found: Tool call manipulation

[!] Found: Memory poisoning

3 vulnerabilities discovered

25
Hands-on Modules
31
Python Labs
25+
Attack Techniques

# Curriculum

01Module 1: What the Hell Is an AI Agent, Actually?theory
02Module 2: Building Your First Agent with LangChain1 lab
03Module 3: Agent Architectures1 lab
04Module 4: NVIDIA NIM Deep Dive1 lab
05Module 5: Direct Prompt Injection3 labs
06Module 6: Indirect Prompt Injection (Environment Poisoning)3 labs
...And 19 more modules covering RAG attacks, MCP exploitation, memory poisoning, agent worms, and more

# Why This Course

⚡

Hands-On Labs

Every module includes runnable Python labs against real NVIDIA NIM APIs. No slides, no theory-only — you exploit, you defend.

🔓

Real Attack Vectors

Prompt injection, tool poisoning, RAG data extraction, supply chain attacks, agent worms, guardrail bypass, and more.

🛡️

Defense-Focused

Each attack lab pairs with a defense lab. Learn to build secure agent architectures from the ground up.

📦

Open Source

All lab code is public on GitHub. Clone, modify, extend. No vendor lock-in, no paywalled content.