
Agentic AI Red Teaming: Hack and Defend AI Agents Hands-On
Published 11/2026
Created by Ing.Seif | Europe Innovation
MP4 | Video: h264, 1920x1080 | Audio: AAC, 44.1 KHz, 2 Ch
Level: All Levels | Genre: eLearning | Language: English | Duration: 15 Lectures ( 1h 16m ) | Size: 1.7 GB
Hijack AI agents with tool abuse, MCP rug pulls and memory poisoning in a safe Kali lab, then prove each fix holds
What you'll learn
⚡ Map an AI agent's attack surface (tools, memory, untrusted input) with the lethal trifecta and the OWASP Top 10 for Agentic Applications
⚡ Build a safe, isolated red-team lab with Kali Linux, VirtualBox and Docker running a deliberately vulnerable AI support agent
⚡ Hijack an agent through a hidden email header and catch the leaked data on a listener you control
⚡ Abuse an agent's tools to cross the backend boundary and reach other customers' data and refunds
⚡ Automate attack regression tests with promptfoo to re-measure any model in minutes
⚡ Write a reproducible red-team report with evidence, honest impact and a verified fix
Requirements
❗ A Windows PC with about 8 GB RAM and internet access
❗ An API key for an OpenAI-compatible model (OpenAI, Anthropic or OpenRouter) with about $1 of credit. That covers every lab in the course
Description
Today's AI models refuse the textbook prompt injection. We tested it: Claude Haiku 5.5 and GPT-6 Luna both resisted the classic "note to the agent" attack in 5 out of 5 runs. Then we attacked the system around the model instead, and every one of the four agentic attacks in this course landed 5 out of 5 times on both models.
That is the real risk with AI agents. The danger is not what the model says. It is what the agent can do: read untrusted tickets, call tools, remember things, and send email on your behalf.
What you will do
In this hands-on course you attack VulnBot, a deliberately vulnerable AI support agent for a fictional online store, running in your own Kali Linux lab. You watch every tool call in a live agent trace, so you see exactly where trust breaks.
✅ Reply-To hijack: make the agent send a customer's private data to an address you control, through a header nobody sees.
✅ Tool abuse: push the agent across the backend boundary to other customers' data and refunds.
✅ MCP rug pull: turn a trusted third-party tool hostile after it was approved.
✅ Memory poisoning: plant a fake "policy" the agent obeys in later sessions.
For every attack you then switch on a backend defense and prove it holds: the attack is blocked, and normal support still works.
How it is organised
Every attack maps to the OWASP Top 10 for Agentic Applications (ASI), so you learn a framework you can use at work, not just tricks. You automate the attacks as regression tests with promptfoo, study real incidents, and finish by writing a red-team report another engineer can reproduce.
What's included
✅ 15 focused lessons, about 1 hour 16 minutes, with no filler
✅ The full VulnBot lab, open source, running only on your machine
✅ Three PDF guides: lab setup and commands, attack and defense map, sample red-team report
✅ An AI role play where you defend your finding to a sceptical product manager
Everything is synthetic: fake customers, fake addresses, fake codes. You practise real attacks safely, in your own virtual machine.
Who this course is for
⭐ Security engineers and pentesters moving into AI and LLM security
⭐ Developers building AI agents who want to see how they get attacked and how to defend them
⭐ AppSec and red-team leads who must test AI agents before launch
⭐ Curious beginners who can follow terminal commands step by step
Homepage
https://www.udemy.com/course/agentic-ai-red-teaming
We need your support!
Make a donation to help us stay online
Bitcoin (BTC)
bc1q08g9d22cxkawsjlf8etuek2pc9n2a3hs4cdrld
Bitcoin Cash (BCH)
qqvwexzhvgauxq2apgc4j0ewvcak6hh6lsnzmvtkem
Ethereum (ETH)
0xb55513D2c91A6e3c497621644ec99e206CDaf239
Litecoin (LTC)
ltc1qt6g2trfv9tjs4qj68sqc4uf0ukvc9jpnsyt59u
USDT (ERC20)
0xb55513D2c91A6e3c497621644ec99e206CDaf239
USDT (TRC20)
TYdPNrz7v1P9riWBWZ317oBgJueheGjATm