审查设计系统与界面一致性
检查设计系统中的组件、状态、令牌与无障碍表现,整理有证据支持的问题、影响和改进方向。
完整提示词
I want a complete audit of the design system and UI consistency with a focus on semantics, components, states, tokens, accessibility and product behavior. Main objective: > Determine where the same concept looks or behaves differently without a justified reason, where the design system allows contradictory patterns and where inconsistency increases cognitive load or implementation risk. This is not: - pixel-perfect uniformity - a ban on local exceptions - "everything must be one component" - automatic consolidation of every similar element - only a Figma audit ## 1. DESIGN SYSTEM INVENTORY - tokens - typography - spacing - color - components - patterns - icons - motion - states - docs ## 2. SOURCE OF TRUTH Code? Design tool? Both? ## 3. TOKEN ## 4. RAW VALUE ## 5. SEMANTIC TOKEN ## 6. THEME ## 7. DARK MODE ## 8. CONTRAST ## 9. TYPOGRAPHY ## 10. SPACING ## 11. RADIUS ## 12. SHADOW ## 13. ICON ## 14. BUTTON ## 15. LINK ## 16. INPUT ## 17. SELECT ## 18. CHECKBOX ## 19. MODAL ## 20. POPOVER ## 21. TOOLTIP ## 22. TABLE ## 23. CARD ## 24. BADGE ## 25. ALERT ## 26. TOAST ## 27. EMPTY STATE ## 28. LOADING ## 29. ERROR ## 30. DISABLED ## 31. HOVER ## 32. FOCUS ## 33. PRESSED ## 34. SELECTED ## 35. DESTRUCTIVE ## 36. SIZE VARIANT ## 37. VISUAL VARIANT ## 38. SEMANTIC VARIANT ## 39. PROP EXPLOSION ## 40. COMPONENT DUPLICATION ## 41. NEAR-DUPLICATE ## 42. WRAPPER ## 43. ONE-OFF ## 44. LOCAL OVERRIDE ## 45. CSS ESCAPE HATCH Not automatically bad. ## 46. COMPONENT API ## 47. COMPOSITION ## 48. ACCESSIBILITY Built into primitive. ## 49. KEYBOARD ## 50. ARIA ## 51. FOCUS MANAGEMENT ## 52. RESPONSIVE ## 53. CONTENT LENGTH ## 54. LOCALIZATION ## 55. THEME OVERRIDE ## 56. BRAND VARIANT ## 57. PRODUCT VARIANT ## 58. DESIGN-CODE DRIFT ## 59. DOCUMENTATION ## 60. STORYBOOK If present. ## 61. VISUAL REGRESSION ## 62. SNAPSHOT ## 63. CHANGE PROCESS ## 64. DEPRECATION ## 65. MIGRATION ## 66. VERSIONING ## 67. ADOPTION ## 68. BYPASS Why teams bypass system? ## 69. UX CONSISTENCY Same destructive action should communicate risk similarly. ## 70. BEHAVIOR CONSISTENCY Not only appearance. ## 71. NAMING ## 72. SEMANTIC COLLISION Two components named differently but same purpose. ## 73. FALSE POSITIVE RULES Do not report: - local exception - multiple button variants - raw CSS - one-off component without actual maintenance/UX consequence. ## 74. EVIDENCE TIERS ```text A - reproduced inconsistency causing user/dev issue B - complete component/code evidence C - strong design-system inconsistency D - suspected consolidation opportunity E - stylistic hardening ``` ## 75. STATUS ```text CONFIRMED LIKELY NOT VERIFIED CONTROLLED NOT APPLICABLE HARDENING ``` ## 76. SEVERITY P0: rare, system inconsistency causes catastrophic wrong action P1: critical semantic component behaves inconsistently in dangerous flow P2: material accessibility/interaction inconsistency P3: maintenance/UX drift P4: visual polish ## 77. FINDING FORMAT ```text ID: Severity: Status: Evidence tier: Pattern/component: Locations: Expected semantic: Variant A: Variant B: User/developer impact: Evidence: Recommended consolidation: Migration risk: ``` ## 78. COMPONENT MATRIX | Semantic role | Components | Behavior same | Visual same | Accessible | |---|---|---|---|---| ## 79. TOKEN MATRIX | Property | Tokenized | Raw overrides | Theme-safe | |---|---|---|---| ## 80. SECOND PASS Search: - buttons - destructive actions - inputs - errors - focus - loading - empty state - modal - table - typography - spacing - responsive variants - dark mode ## 81. FINAL QUALITY GATE Confirm: - tokens - components - semantics - states - accessibility - behavior - variants - documentation - design-code parity - migration ## 82. OUTPUT `DESIGN_SYSTEM_CONSISTENCY_AUDIT.md` # FINAL RULE A design system is not good because everything looks the same. It is good when: ```text the same concept has a predictable meaning and behavior + exceptions are intentional + the implementation stays maintainable ```
分类与标签
设计创意、设计系统、界面一致性、无障碍审查