Architect perspective on Round 1 automation classification: the current allowlist approach is structurally correct but needs an extended trigger vocabulary and a third 'authority' axis beyond execution_mode + security_relevance.
Architect perspective on Round 3 automation-persistence analysis: a lightweight execution_chains collection with versioned snapshots that gives chains stable identity and chain-level temporal tracking without duplicating entity-level temporal machinery.
Architect perspective on Round 2 execution-flow analysis: six backward-compatible changes to restore full BR-to-SI-to-SP-to-resource execution provenance without adding a new entity_type.