What problem does it solve? Engineers often need to know whether code, architecture, or design is genuinely good — not just whether it works. This Skill provides calibrated qualitative judgment across five dimensions, replacing vague gut feelings with articulated, defensible reasoning about craftsmanship. ## Core Features & Use Cases - Quick Judgment: Delivers a prose verdict plus the single highest-leverage fix for single files, snippets, or simple "is this good" questions. - Deep Review: Performs dimension-by-dimension analysis with optional 1-5 scores for large codebases, architecture documents, or PRDs, with an optional reviewer agent for targets spanning five or more files. - Guide & Compare: Applies taste principles during code generation and refactoring with codebase pattern adaptation, and compares approaches while labeling taste judgments versus objective tradeoffs. - Use Case: Paste a newly written module and ask whether it is well-designed; receive a verdict on naming, cognitive load, error handling, and consistency, plus the one change that would most improve it. ## Quick Start Ask the assistant to review the taste of a specific file or code snippet and tell you whether it is well-designed.
