> Hello,
>
> I would like to suggest two risks for the Agentic Identity Risk Universe. Please treat this as feedback only, not as a contribution under the group's participation terms.
>
> 1. Under "Delegation chain risks", a principal or an intermediate agent withdraws authority, but a subagent keeps acting because the verifier checks only the credential closest to the action. A public test fixture checks that revoking the root or middle delegation invalidates a chain of three delegations even when descendant delegations remain unrevoked.
> https://github.com/Agent-Authority-Conformance/aps-conformance-suite/tree/m…
>
> 2. Under "Stale access permissions", a counterparty keeps relying on a cached authorization decision after the authority behind it was revoked, so the agent can keep acting until the cache expires.
>
> The fixture is from a conformance lab I maintain. Its results come from its authors and concern fixed inputs. They do not establish whether any deployed system has either risk.
>
> Would the group want these as candidate rows for the next revision? Does feedback require participation in the discussion group?
>
> Tymofii Pidlisnyi (Agent Passport System)