What problem does it solve? UI reviews often produce subjective opinions, unsupported nitpicks, or sweeping redesigns that ignore the existing design system. This Skill audits one coherent product surface against the design evidence that actually governs it, keeps only findings proven by contract, runtime path, and a deterministic correction, and turns selected findings into self-contained implementation plans another agent can execute. ## Core Features & Use Cases - Evidence-gated auditing: Traces the rendered path through routes, layouts, components, variants, and tokens, then requires contract, runtime, and correction proofs before any finding is reported. - Design-system reconstruction: Reads DESIGN.md and surface-local documentation, resolves token aliases and variants, and records governing owners and documented exceptions. - Plan generation for handoff: Writes one self-contained plan per selected change using a structured template covering evidence chain, reuse, scope, validation, and stop conditions, while never modifying product source. - Use Case: Ask it to review your app's settings surface; it reports at most three verified findings ranked by confidence and impact, then drafts an implementation-ready plan only for the finding you choose. ## Quick Start Use the improve-ui skill to audit the primary product surface, report supported UI problems, and recommend what to improve first without changing any code.
