productionos-self-eval

Automated self-evaluation scores recent artifacts against standardized criteria to identify quality, correctness, dependency, and completeness gaps before delivery. It evaluates the latest session or diff to guide quality improvement.

Adversarial validation in deep mode hardens self-evaluation by aggressively testing recent work for honesty concerns and hidden quality issues. It provides a stricter assessment of artifacts beyond standard code review checks.

Yes, conditional self-evaluation results can be routed into an optional healing repair loop to fix identified issues. This prevents low-quality output by automatically addressing detected quality gaps.

Self-evaluation enforces read-only analysis by scoring recent diffs or explicit paths without modifying them. It strictly assesses quality and completeness gaps to generate repair guidance rather than applying changes.

Running an automated self-evaluation workflow is the best way to check completeness and dependencies. It scores recent work against standardized criteria to identify missing elements and guide quality improvements before delivery.

Use deep mode self-evaluation on a session when you need adversarial validation to harden quality assessment. It is ideal for catching honesty concerns and hidden correctness gaps that standard evaluation might miss.