What problem does it solve? Before implementation begins, specifications often contain gaps, contradictions, or infeasible approaches that cause rework later. This Skill systematically critiques a story's spec.md against its requirements.json, scoring five quality dimensions and producing a blocking gate verdict (APPROVED, NEEDS_REVISION, or BLOCKED) so weak specs are caught early. ## Core Features & Use Cases - Five-Dimension Scoring: Evaluates accuracy, completeness, consistency, feasibility, and alignment with weighted checks and 1-5 scores per dimension. - Structured Issue Reports: Generates critique.json with severity-tagged issues (HIGH/MEDIUM/LOW), locations, fix suggestions, and auto-fixable flags. - Pipeline Gate Integration: Acts as a blocking verification gate between the spec and plan phases, routing NEEDS_REVISION specs back to spec-write and escalating BLOCKED specs to the architect. - Use Case: After drafting a specification for STORY-42, run the critique to discover that two functional requirements lack test scenarios, receive an auto-fixable suggestion, and return the spec for revision before planning. ## Quick Start Ask the agent to critique the specification for story STORY-42 and report the verdict with any issues found.
