self-critique

To verify policy compliance before returning a draft, trigger a self-critique gate that checks the draft against frozen constitutional principles, logging the verdict and revision count to ensure alignment before shipping.

A constitutional check is a pre-return verification mechanism that evaluates draft outputs against frozen principles to prevent policy drift and ensure unsafe or non-compliant responses are corrected before delivery.

You audit and revise drafts using a two-phase loop where the draft is checked against constitutional principles, revised if necessary, and escalated after two unsuccessful rounds of failing the self-critique.

Yes, the self-critique workflow integrates with related governance skills like injection-defense, audit, and self-knowledge to maintain consistency, safety, and policy compliance across the agent system.

When an output fails constitutional checks after two unsuccessful revision rounds, the self-critique mechanism escalates the draft to prevent shipping a non-compliant or unsafe response.