Securing Agentic AI Systems — Course Waitlist
Securing Agentic AI Systems
Hands-on, framework-grounded, red-team-tested.
Practical playbooks built on OWASP, MITRE ATLAS, and NIST guidance — for security engineers and CTOs deploying agents in production.
What this course is
Most AI security training covers cloud misconfigs and classic AppSec. This course covers what’s actually different about agentic systems: the new attack surface that emerges when an LLM controls tools, makes decisions, and calls external APIs.
Four weeks of live sessions, structured labs, and direct review of your own architecture. Every module maps directly to something you can implement the week you learn it.
Not theory. Not FUD. Playbooks you ship.
What you will leave with
A working threat model for agentic AI
Map the full attack surface of your agentic system across all 10 OWASP Agentic risks — prompt injection, tool misuse, privilege escalation, and beyond.
Red team methodology for AI systems
Structured 5-phase approach to adversarial testing of LLM-powered systems. Run it yourself or hand it to an external team.
MCP and tool security architecture
Tool poisoning, meta-context injection, cross-server attacks. How they work, what they break, and exactly how to mitigate them.
Production control library
Human approval gates, least-privilege tooling, mcp-scan, tool description hashing, per-server namespacing. Deployable this sprint.
Who this is for
- Security engineers tasked with reviewing, hardening, or red-teaming an LLM or agentic system
- CTOs and engineering leads who need to understand the risk surface of AI they are deploying or have deployed
- Platform and infrastructure teams running agentic workloads on Kubernetes or cloud infrastructure
- Compliance and risk officers extending SOC2, ISO 27001, or GDPR programs to cover agentic AI
If you are trying to understand what changes when the AI controls the tools — not just the model — this course is for you.
Curriculum overview
Week 1 — The Agentic Attack Surface OWASP Agentic Top 10 in depth. ASI01–ASI10 mapped to real incidents with detection and mitigation strategies.
Week 2 — MCP Security and Tool Architecture MCP attack patterns, tool poisoning, cross-server attacks. Secure tool design and least-privilege architecture.
Week 3 — Red Teaming Agentic Systems 5-phase red team methodology. Promptfoo, mcp-scan, automated and manual test patterns. Building your red team playbook.
Week 4 — Production Hardening and Incident Response Control implementation, monitoring for AI decisions, audit logging, and response playbooks for agentic incidents.
Join the waitlist
Waitlist members are notified first when enrollment opens and receive early-access pricing. No commitment required.
No spam. One email when enrollment opens.
Amine Raji, PhD · CISSP · aminrj.com