Cybersecurity EngineerAdvanced6 modules~3.0 hours hands-on3 learning modes
AI, LLM & Agent Security
Constrain prompt injection, untrusted retrieval, tools, memory, identity, output, and approval boundaries in a local deterministic agent gateway.
// What you'll be able to do- Instructions inside content do not gain authority
- Every action is typed, scoped, authorized, and bounded
- Persist only reviewed facts and validate every downstream use
- Human approval is a security boundary only when it is informed
6 modules, built to run
Every module ships runnable code. Expand any one to see what's inside.
00IntroductionThe model is a probabilistic component inside a security system30 min
Objective — The model is a probabilistic component inside a security system
Start moduleFree preview
01Prompt Injection & RetrievalInstructions inside content do not gain authority30 min
Objective — Instructions inside content do not gain authority
02Tools, Identity & Excessive AgencyEvery action is typed, scoped, authorized, and bounded30 min
Objective — Every action is typed, scoped, authorized, and bounded
03Memory, Output & Supply ChainPersist only reviewed facts and validate every downstream use30 min
Objective — Persist only reviewed facts and validate every downstream use
04Approval, Monitoring & EvalsHuman approval is a security boundary only when it is informed30 min
Objective — Human approval is a security boundary only when it is informed
05Pattern SynthesisBuild and verify a deterministic agent gateway30 min
Objective — Build and verify a deterministic agent gateway
Ready to start AI, LLM & Agent Security?
Start with the free Module 0, then unlock the full course on a Pro plan.