Skip to content
From LLMs to Secure Agents Robot MascotFrom LLMs to Secure Agents
Start ReadingGitHub

Agentic AI and its security

A comprehensive, sequential guide to the architecture, operation, risks, and defenses of agentic systems.

Start with prerequisitesBrowse the index
The guide progresses from a language model to a complete agentic system, then revisits that system with security boundaries and runtime controls

One system, studied in order

Agentic AI combines models with software components that hold context, call tools, preserve state, and produce side effects. The guide treats that whole system as the subject.

Build the functional model

Begin with software foundations, agent behavior, orchestration patterns, and the components used in complete agentic systems.

Study security in context

Map threats, boundaries, controls, and tests back to the same components and workflows instead of learning them as an isolated checklist.

Chapters are grounded in official specifications and primary research, with local diagrams and minimal runnable examples where they help.