AI Beyond Prompts: Agentic AI for Cloud Security

This session covers how agentic AI differs from chat-based AI and how cloud security teams can apply it to their workflows. Attendees will learn the four building blocks of GitHub Copilot CLI —skills, agents, sessions, and instructions — and how each maps to security operations. Two demonstrations show the concepts in practice: building a Terraform tag compliance skill from scratch, and using a multi-skill agent to research CIS S3 benchmark controls, generate compliant and non-compliant infrastructure, and validate detection coverage. The session assumes no prior AI tooling experience.

Who Should Attend?

• Cloud security engineers and architects
• DevSecOps engineers managing IaC pipelines
• Security team leads evaluating AI tooling for their teams
• Anyone curious about applying agentic AI to security workflows

Learning Objectives:

• Define agentic AI vs chat-based AI
• Identify the four Copilot CLI primitives
• Write a reusable SKILL.md
• Construct an AGENT.md orchestrating multiple skills
• Apply the session feedback loop for improvement
• Generate, deploy, and validate Terraform against CIS benchmarks