12  Agentic Safety × Security

13 Agentic Safety × Security

Flagship chapter — stub. The synthesis the whole book points at: where autonomy (safety) meets tool-use and external action (security).

Chronological deep-dive: single-agent tool-use risk → guardrail architectures (input / planning / action / output / systemic) → prompt-injection and permission-scoping defenses → multi-agent trust and emergent-behavior control → evaluation of agentic safety. Feeds the evaluation and guardrails review papers.