gsd-eval-review

To audit AI evaluation coverage, you can retroactively review the completed phase and score each evaluation dimension as COVERED, PARTIAL, or MISSING to identify gaps in adherence to the evaluation strategy.

Generating a remediation plan involves scoring evaluation dimensions across completed AI phases to surface gaps, which automatically produces an EVAL-REVIEW.md file containing the actionable remediation plan and concise verdicts.

Retroactive AI phase evaluation improves traceability by verifying adherence to the evaluation strategy, scoring coverage across dimensions, and generating a documented gap list with a remediation plan for future reference.

You do not need to specify a phase number to run the retroactive evaluation workflow, as the phase number is optional for generating the EVAL-REVIEW.md with the verdict, gap list, and remediation plan.

Yes, you can use this audit to score partial evaluation coverage in AI phases by evaluating each dimension and assigning a score of COVERED, PARTIAL, or MISSING to quantify gaps accurately.