Skip to content
AIGP

RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI — 24. Evaluation Lifecycle

AIGP Specification › RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI › 24. Evaluation Lifecycle

← 23. Conflict Resolution · Section index · 25. Failure Modes →

24. Evaluation Lifecycle

AIGP-SGL runtime evaluation follows this lifecycle:

1. Receive proposed action.
2. Resolve subject, action, object, and context.
3. Load active Governance Intent.
4. Load applicable operational envelope.
5. Load applicable SGL constraints.
6. Validate authority.
7. Validate D-DNA chain and prior state.
8. Evaluate runtime context.
9. Apply conflict resolution.
10. Produce decision: permit, deny, inhibit, modify, escalate, degrade, pause, return.
11. Execute required obligations.
12. Record D-DNA event.
13. Emit explanation if required.

← 23. Conflict Resolution · Section index · 25. Failure Modes →