This course is designed for security practitioners, penetration testers, red teamers, developers, and aspiring AI systems engineers who want to understand how agentic AI systems work and where they can be abused. You will learn the core concepts behind agentic AI, including agents, agent loops, tool use, autonomy levels, trust boundaries, and the relationship between agents, tools, and application workflows.
The course also explores abuse patterns in agentic pipelines and multi-agent systems. In this course, you will learn how to examine multi-agent workflows, agent chains, instruction propagation, and agent chain injection, culminating in a practical lab where you break an agent chain and observe how poisoned context can move across agents and lead to unauthorized actions.
