Several frameworks adapt threat modeling to agents by adding layers for models, memory, tools, and agent to agent communication. They work as checklists once you know the basics, and worked audits show how practitioners apply them to real products.
Read in this order
- OfficialAgentic AI: Threats and Mitigations (OWASP GenAI Security Project)Start here: a catalog of agent-specific threats, each paired with mitigations.
- VideoAI Agent Security - MIT 6.566 Computer Systems Security, Spring 2026A systems-security lecture with code and chapters on isolation. 85 min.
- ArticleUsing threat modeling and prompt injection to audit Comet (Trail of Bits)A real audit of an AI browser, showing the method in practice.