What problem does it solve? Design decisions scattered across a codebase become inconsistent and get re-litigated every session. This Skill extracts the implicit system from existing code, names it by role, and records decisions with their reasons so future work stays coherent. ## Core Features & Use Cases - Token extraction and consolidation: Inventory real values in use, group near-duplicates, and migrate deviations into named scales for color, spacing, typography, radius, shadow, and motion. - Role-based color and dark mode: Define colors by role (surface, text, accent, focus) instead of hex lists, with a three-state dark mode strategy that redefines values without redeclaring structure. - Fluid scales and DESIGN.md: Use clamp()-based fluid typography and spacing, and maintain a DESIGN.md file capturing direction, tokens, decisions with reasons, and out-of-scope items. - Use Case: When a codebase has 14px, 15px, and 0.9rem scattered everywhere, use this Skill to consolidate them into a named type scale and document why each step exists. ## Quick Start Ask the AI to extract the design tokens from this codebase and create a DESIGN.md documenting the scales, colors, and the reasons behind each decision.
