What problem does it solve? Reviewing an AI Skill package by intuition alone leads to inconsistent judgments: vague triggers, missing output contracts, and unverified behavior slip through. This Skill provides a structured six-dimension audit (trigger boundaries, workflow, output contract, verification, safety, maintainability) with evidence-based PASS/PARTIAL/BLOCKED verdicts. ## Core Features & Use Cases - Six-Dimension Review: Assesses trigger specificity, workflow ordering, output contracts, verification coverage, safety boundaries, and maintainability, recording facts, inferences, and missing evidence separately. - Deterministic Preflight Script: Runs a standard-library-only Python script that checks frontmatter, section signals, UI metadata, and non-empty examples across one Skill or an entire repository. - Prioritized Findings: Outputs P0/P1/P2 ranked minimal fixes with evidence paths, residual risks, and explicitly out-of-scope items. - Use Case: Before publishing a modified Skill, run the gate to confirm its description states non-applicable scenarios, its examples are non-empty, and its regression checks actually execute. ## Quick Start Use the skill-quality-gate to review the natural-rewrite Skill package and report problems without modifying any files.
