1A Common Language for Agent IdentityAI agents act for people and for other systems — but without a shared, verifiable way to say who an agent is and what it may do, trust stops at every boundary.Advanced 2Intent-Bound AuthorizationWhy an agent should be granted permission for a specific task it was asked to do — not a standing licence to act as you forever.Advanced 3Federated Delegation: One Agent Acting for AnotherHow an agent can safely call another agent on your behalf across a trust boundary — carrying your authority without becoming you.Advanced 4Authorizing Time: Order, Staleness & ReplayAuthority is not only about what and who — it is about when, and an agent that ignores time can be tricked with permissions that were once perfectly valid.Advanced 5An Evidence Budget for AgentsProof should cost what the action is worth — a routine read needs almost none, moving money needs a lot, and spending evidence in between is a deliberate design choice.Advanced 6Triaging Agent DeviationsWhen an agent steps outside its authority, the difference between a contained incident and a disaster is what you do in the first moments — and whether you planned it.Advanced