Agentic AI security training

Twenty modules derived from the repo's security-best-practices-series. Content is synced at site build for static HTML and search. Edit the Markdown in the repo, then run node scripts/sync-security-training-modules.mjs from website/ (also runs on prebuild).

← Security overview · Defense in depth (full guide)

1. 1.Supply Chain Security: Why Pinning Versions and Running Your Own Mirror Registry MattersExplain why the software supply chain is a primary risk for agentic AI and tool-calling platforms.
2. 2.Building Golden Images: Automated Scanning, Hardening, and Distroless PipelinesExplain why minimal (e.g. distroless) images and read-only root filesystems reduce container blast radius.
3. 3.Cluster Admission Control: Enforcing Image Signing and Policy at Deploy TimeDescribe the role of admission controllers in preventing mis-scoped workloads from running.
4. 4.Principle of Least Privilege: Scoped Identities and Limiting Blast RadiusApply least privilege to Kubernetes identities (ServiceAccounts, Roles, bindings).
5. 5.Zero Trust Fundamentals: Assume Compromise and Verify EverythingState Zero Trust principles in the context of autonomous agents and external tools.
6. 6.Advanced Zero Trust: Multi-Sig Vault, HSM, Tamper-Proof Logging, and Cryptographic ProvenanceCompare static vs dynamic secrets and justify short TTLs for machine identities.
7. 7.RBAC, mTLS, and Istio Service Mesh: Network-Level Zero TrustExplain mutual TLS and service identity for east-west traffic in Kubernetes.
8. 8.Sandboxing Options and Trade-offs: Kata, gVisor, Seatbelt, Docker, and Cloudflare WorkersCompare isolation technologies (VM-backed runtimes, user-space kernels, OS sandboxes).
9. 9.MCP Runtime Protection: Panguard, ATR Rules, and Agentic Threat MitigationExplain synchronous policy enforcement for tool and API calls in agent architectures.
10. 10.Data Classification and PII Redaction: Never Let Sensitive Data Hit LogsDistinguish data classification from redaction and logging policy.
11. 11.Model Integrity: Verifying Weights Before InferenceExplain why model artifacts need integrity checks beyond container image scanning.
12. 12.Runtime Monitoring and Observability: Falco, Wazuh, Prometheus, and Merkle MetricsLayer host-level detection, SIEM correlation, metrics, and tracing for AI platforms.
13. 13.Automated Response and Containment: Falco + Talon Quarantine, Panguard BlockingMap alert confidence to automated vs manual response actions.
14. 14.Incident Response and Recovery: PICERL, WORM Audits, and Tested BackupsUse a structured incident lifecycle (e.g. prepare → identify → contain → recover).
15. 15.GPU and Resource Protection: Preventing Rogue Agent Denial-of-ServiceApply quotas, limits, and scheduling policies to protect shared GPU pools.
16. 16.Workstation and Local Development Security: Same Posture EverywhereExtend production security expectations to developer laptops and CI runners.
17. 17.Production Deployment: One-Command Secure Full StackAssemble a repeatable secure rollout checklist for complex stacks.
18. 18.Threat Modeling with STRIDE for Agentic AI SystemsApply STRIDE categories to agent identity, tools, memory, and orchestration.
19. 19.OWASP Agentic Top 10: Mapping Risks to Architectural ControlsNavigate the OWASP Agentic risk list and map items to layered controls.
20. 20.Quarterly Security Review Checklist: Keeping Defense-in-Depth AliveRun a periodic defense-in-depth review across supply chain, runtime, and data.