AI-powered agents are becoming core to modern software systems, yet security practices have not kept pace. This masterclass bridges that gap by teaching professionals how to build secure, resilient, and observable AI agent systems.
Across two days, participants will learn how agents operate using LLMs, tools, skills, and MCP servers, and how these components expand the attack surface. The course covers secure RAG pipelines, agent memory systems, and Skill-based architectures, highlighting vulnerabilities such as prompt injection, data poisoning, and cross-system contamination.
Participants will then shift to MCP security design, including approvals, elicitations, OAuth 2.1 authorization, and secure tool orchestration. Offensive modules demonstrate real-world attacks such as excessive agency exploitation, tool misuse, and MCP shadowing attacks.
On the defensive side, attendees implement guardrails (Llama Guard, NeMo Guardrails, Guardrails AI), sandboxing (microVMs, gVisor, E2B/Daytona), identity systems (SPIFFE/SPIRE), and runtime monitoring with observability tools like Gryph, Langfuse, and Arize Phoenix.
The course concludes with supply chain security for MCP ecosystems, continuous evaluation, and production-grade detection strategies.
.png)
.png)
Koushik M.
"Exceptional Hands-On Security Learning Platform"
Varunsainadh K.
"Practical Security Training with Real-World Labs"
Gaël Z.
"A new generation platform showing both attacks and remediations"
Nanak S.
"Best resource to learn for appsec and product security"
.svg)
.svg)
.svg)
.svg)