Free course material
Put guardrails where agents take action.
Agent Security
A practical course on securing tool-using and autonomous systems from first principles through production operations.
What this course is about
A practical path from understanding to application.
Agent security requires more than a safer prompt. This course maps the attack surface of agentic systems, then works through identity, tool authorization, prompt injection, data exfiltration, supply-chain risk, monitoring, incident response, and the controls needed when an agent can affect real systems.
Covered material
What learners will work through.
- Agent threat modeling and the expanded attack surface
- Prompt injection, indirect injection, data exfiltration, and unsafe outputs
- Tool authorization, sandboxing, secrets, isolation, and least privilege
- Identity, delegation, auditability, monitoring, and incident response
- Supply-chain, model, retrieval, memory, and multi-agent security concerns
- Security testing, red teaming, operational controls, and production readiness
Who it is for
Meet learners where they are.
Security practitioners, engineers, platform teams, architects, risk owners, and leaders who need to understand and govern the risks introduced by tool-using AI systems.
Adapt this course
Bring the material to your organization.
We can adapt the level, examples, exercises, and pace for executives, non-technical groups, technical teams, or mixed audiences. Sessions combine theory with practical and coding work where it fits, led by educators who meet with your group.
Book a custom training conversationRelated One+i service
AI governance & technical leadership
Connect open learning with the strategy, delivery, and adoption support around it.