Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Pick a wine that flatters tonight's meal — at your budget, from what's actually available — without the sommelier mystique. Use when asked what wine goes with [dish], help me pick a wine, what should I drink with dinner, or recommend a bottle for. Produces a couple of specific bottle styles (not just 'a red'), why each works with the dish, a budget-tier pick, an easy-to-find fallback, and a non-alcoholic option — with a plain reason you can remember next time.
.claude/skills/mohitagw15856-wine-pairing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 65% | 0% |
Pairing isn't a secret code — it's a few rules about matching weight, acidity, and intensity, plus what you can actually buy nearby. This gives you specific styles to ask for (or grab off the shelf), a reason each works, and options across price — so you walk in confident instead of grabbing the bottle with the nicest label.
Ask for these if not provided:
🍷 Style/grape] — works because weight/acidity/etc.]. Ask for: example]. 🍷 Style/grape] — works because …].
Budget pick: style + rough price] · Step up: style]. Easy to find anywhere: safe fallback]. No/low-alcohol: option].
Remember this: the one rule that made it work].
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 14,011 | 14,866 | +6% | 1 | 1 | 0% | 2,341 | 2,591 | +11% | 0 | 0 | — |
case-02 | pass→pass | 20,698 | 15,501 | -25% | 1 | 1 | 0% | 2,607 | 2,584 | -1% | 0 | 0 | — |
case-03 | pass→pass | 12,688 | 14,962 | +18% | 1 | 1 | 0% | 2,122 | 2,566 | +21% | 0 | 0 | — |
case-04 | fail→pass | 13,344 | 15,807 | +18% | 1 | 1 | 0% | 2,133 | 2,642 | +24% | 0 | 0 | — |
case-05 | pass→pass | 18,946 | 18,717 | -1% | 1 | 1 | 0% | 2,170 | 3,048 | +40% | 0 | 0 | — |
case-06 | fail→pass | 19,290 | 14,887 | -23% | 1 | 1 | 0% | 2,304 | 2,491 | +8% | 0 | 0 | — |
case-07 | pass→pass | 9,974 | 11,888 | +19% | 1 | 1 | 0% | 1,550 | 2,472 | +59% | 0 | 0 | — |
case-08 | pass→pass | 17,531 | 14,998 | -14% | 1 | 1 | 0% | 2,144 | 2,595 | +21% | 0 | 0 | — |
case-09 | pass→pass | 13,827 | 14,931 | +8% | 1 | 1 | 0% | 2,171 | 2,514 | +16% | 0 | 0 | — |
case-10 | fail→pass | 20,258 | 16,825 | -17% | 1 | 1 | 0% | 2,440 | 2,545 | +4% | 0 | 0 | — |
case-11 | pass→pass | 16,935 | 9,080 | -46% | 1 | 1 | 0% | 1,824 | 2,403 | +32% | 0 | 0 | — |
case-12 | fail→pass | 14,688 | 10,949 | -25% | 1 | 1 | 0% | 1,978 | 2,673 | +35% | 0 | 0 | — |
case-13 | pass→pass | 16,143 | 17,478 | +8% | 1 | 1 | 0% | 2,415 | 2,466 | +2% | 0 | 0 | — |
case-14 | pass→pass | 18,331 | 10,808 | -41% | 1 | 1 | 0% | 2,092 | 2,631 | +26% | 0 | 0 | — |
case-15 | pass→pass | 11,463 | 16,168 | +41% | 1 | 1 | 0% | 1,848 | 2,535 | +37% | 0 | 0 | — |
case-16 | fail→pass | 16,570 | 17,310 | +4% | 1 | 1 | 0% | 1,714 | 2,826 | +65% | 0 | 0 | — |
case-17 | fail→fail | 17,873 | 12,284 | -31% | 1 | 1 | 0% | 1,952 | 2,791 | +43% | 0 | 0 | — |
case-18 | fail→pass | 18,581 | 15,044 | -19% | 1 | 1 | 0% | 2,035 | 2,373 | +17% | 0 | 0 | — |
case-19 | pass→pass | 16,806 | 14,001 | -17% | 1 | 1 | 0% | 1,901 | 2,235 | +18% | 0 | 0 | — |
case-20 | pass→pass | 17,819 | 16,793 | -6% | 1 | 1 | 0% | 2,143 | 2,748 | +28% | 0 | 0 | — |
case-21 | pass→pass | 12,686 | 16,899 | +33% | 1 | 1 | 0% | 2,154 | 2,747 | +28% | 0 | 0 | — |
case-22 | pass→pass | 21,918 | 19,992 | -9% | 1 | 1 | 0% | 2,897 | 3,482 | +20% | 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. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 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.