What problem does it solve? Draft planning artifacts like implementation plans, slice specs, and PR packages often contain vague exit criteria, scope creep, or missing risk notes, and re-running the authoring command to fix them is expensive. This Skill runs a single structured critique pass against a fixed rubric and emits a revised draft, so draft quality improves without restarting the upstream workflow. ## Core Features & Use Cases - Locked per-artifact rubrics: Applies a fixed 7-criterion rubric for IMPLEMENTATION_PLAN.md, SLICES/*.md, or PR_PACKAGE.md, scoring each criterion PASS, FAIL, or WEAK with reasoning. - Conservative revision: Rewrites only FAIL and WEAK items fixable from existing context, preserves passing content verbatim, and defers judgment calls to a Not applied section. - Workflow routing: Detects unsupported artifacts and emits NO_OP_TRACE with routing to the right command (review-hard, sync-task-state, direction-adjust). - Use Case: After drafting an implementation plan whose slices lack verifiable exit criteria, run this Skill to get a criterion-by-criterion critique plus a revised plan with concrete exit criteria, while deferring the question of splitting slice 03 to the user. ## Quick Start Ask the AI to run self-critique-and-revise on the IMPLEMENTATION_PLAN.md in the active task folder and show the critique plus revised draft for approval.
