Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Upgrade an existing website or app to premium quality without breaking functionality — audit the current design, identify generic/AI tells, then apply taste and system rules surgically. Use when the user wants to improve, modernize, polish, or "make better" an existing UI/codebase.
.claude/skills/plugin87-redesign/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -16% | 0% |
Audit-first redesign that preserves behavior.
workflows/redesign-audit.md (the full Scan → Diagnose → Direct → Apply → Verify sequence + Output Completeness).frameworks/ adapter), token reality, components, what must be preserved.taste/design-taste.md) + the review rubric (design-review skill). Produce a prioritized findings table.apply-aesthetic skill) that fits the brand.design-review + a11y-audit; smoke-test every previously working flow; dark mode + responsive spot-check. Run scripts/validate_contrast.py on the token source and scripts/lint_hardcodes.py over the changed code.Never sacrifice a working feature for aesthetics. Never ship a brand color that fails contrast. Never remove existing accessibility affordances. Deliver complete files — no placeholders.
Consolidate to ONE shared token theme and make every page consume it — a redesign that leaves different pages on different palettes has failed. Replace per-page/ad-hoc colors with semantic tokens; verify with scripts/lint_hardcodes.py that no page reintroduces off-theme values. Theme switches must come from the single token source, not page edits (CLAUDE.md → Single-Theme Consistency).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 32,375 | 3,742 | -88% | 1 | 1 | 0% | 3,631 | 558 | -85% | 0 | 0 | — |
case-08 | pass→pass | 13,726 | 14,848 | +8% | 1 | 1 | 0% | 1,884 | 2,498 | +33% | 0 | 0 | — |
case-02 | fail→fail | 25,922 | 2,778 | -89% | 1 | 1 | 0% | 5,005 | 765 | -85% | 0 | 0 | — |
case-03 | fail→fail | 27,745 | 5,687 | -80% | 1 | 1 | 0% | 6,191 | 1,192 | -81% | 0 | 0 | — |
case-04 | pass→pass | 17,160 | 26,163 | +52% | 1 | 1 | 0% | 3,022 | 5,151 | +70% | 0 | 0 | — |
case-05 | pass→pass | 20,926 | 18,872 | -10% | 1 | 1 | 0% | 3,926 | 4,347 | +11% | 0 | 0 | — |
case-06 | pass→pass | 15,220 | 16,151 | +6% | 1 | 1 | 0% | 3,286 | 3,578 | +9% | 0 | 0 | — |
case-07 | pass→pass | 16,076 | 11,502 | -28% | 1 | 1 | 0% | 2,257 | 1,962 | -13% | 0 | 0 | — |
case-09 | fail→pass | 12,577 | 13,680 | +9% | 1 | 1 | 0% | 1,856 | 2,375 | +28% | 0 | 0 | — |
case-10 | pass→pass | 13,919 | 13,421 | -4% | 1 | 1 | 0% | 2,467 | 2,700 | +9% | 0 | 0 | — |
case-11 | fail→pass | 16,088 | 3,396 | -79% | 1 | 1 | 0% | 2,414 | 826 | -66% | 0 | 0 | — |
case-12 | fail→pass | 19,943 | 14,080 | -29% | 1 | 1 | 0% | 3,376 | 2,935 | -13% | 0 | 0 | — |
case-17 | pass→pass | 10,888 | 9,886 | -9% | 1 | 1 | 0% | 1,768 | 2,068 | +17% | 0 | 0 | — |
case-13 | fail→pass | 21,909 | 12,525 | -43% | 1 | 1 | 0% | 3,265 | 2,387 | -27% | 0 | 0 | — |
case-14 | fail→fail | 10,801 | 6,193 | -43% | 1 | 1 | 0% | 1,643 | 1,262 | -23% | 0 | 0 | — |
case-15 | fail→pass | 14,696 | 9,871 | -33% | 1 | 1 | 0% | 2,159 | 1,821 | -16% | 0 | 0 | — |
case-16 | fail→pass | 13,585 | 2,482 | -82% | 1 | 1 | 0% | 2,039 | 737 | -64% | 0 | 0 | — |
case-18 | pass→pass | 13,841 | 14,638 | +6% | 1 | 1 | 0% | 1,992 | 2,767 | +39% | 0 | 0 | — |
case-19 | pass→pass | 14,783 | 9,679 | -35% | 1 | 1 | 0% | 2,097 | 1,872 | -11% | 0 | 0 | — |
case-20 | pass→pass | 11,410 | 7,423 | -35% | 1 | 1 | 0% | 1,746 | 1,703 | -2% | 0 | 0 | — |
case-21 | pass→pass | 16,764 | 9,961 | -41% | 1 | 1 | 0% | 2,480 | 1,989 | -20% | 0 | 0 | — |
case-22 | pass→pass | 11,778 | 11,352 | -4% | 1 | 1 | 0% | 1,654 | 1,866 | +13% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.