Use when a user asks to review, interpret, or audit an AI agent skill (SKILL.md plus bundled scripts/references/assets) for capabilities, triggering behavior, tool/command usage, safety & privacy risk, supply-chain provenance, quality gaps, and improvement recommendations; also use when validating a skill before installing or deploying it.