Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Run Shopify stores for Taiwan market — localization (NT$ pricing, Traditional Chinese, TW address format), payment apps (ECPay / NewebPay / TapPay Shopify apps), shipping apps (CVS / 黑貓), and e-invoice integration. Use when setting up Shopify for TW, choosing TW payment/shipping apps, or adapting a global Shopify theme for TW. Do NOT use for general (non-TW) Shopify dev. STATUS: SKELETON — body pending.
.claude/skills/asgard-ai-platform-tw-ecom-dtc-shopify-localization/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-01 | ✓→✓ | = Same ✓ | -2% | 0% |
| case-02 | ✓→✓ | = Same ✓ | -4% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 24% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 11% | 0% |
> STATUS: SKELETON — body pending.
TODO: Shopify market/region config, app-based extension model (TW via marketplace apps, not native), why no official mcp-shopify yet for TW context.
TODO: when Shopify beats Shopline/91APP for TW (brand-led international expansion, heavy theme customization need).
TODO: app selection, localization checklist, e-invoice hook.
TODO: 5-6 pitfalls (theme i18n gaps, app double-charging, payment app settlement differences, CVS shipping quirks).
TODO.
TODO.
tw-ecom-channel-strategytw-ecom-payment-newebpaytw-ecom-logistics-cvs_Last verified: 2026-04_
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 24,558 | 31,888 | +30% | 1 | 1 | 0% | 3,530 | 3,453 | -2% | 0 | 0 | — |
case-02 | pass→pass | 13,849 | 12,067 | -13% | 1 | 1 | 0% | 2,273 | 2,180 | -4% | 0 | 0 | — |
case-03 | pass→pass | 20,011 | 26,878 | +34% | 1 | 1 | 0% | 3,332 | 4,140 | +24% | 0 | 0 | — |
case-04 | pass→pass | 13,839 | 14,977 | +8% | 1 | 1 | 0% | 2,173 | 2,411 | +11% | 0 | 0 | — |
case-05 | pass→pass | 18,697 | 17,954 | -4% | 1 | 1 | 0% | 3,012 | 3,323 | +10% | 0 | 0 | — |
case-06 | pass→pass | 16,623 | 27,811 | +67% | 1 | 1 | 0% | 2,744 | 4,340 | +58% | 0 | 0 | — |
case-07 | fail→fail | 18,331 | 13,114 | -28% | 1 | 1 | 0% | 2,550 | 2,302 | -10% | 0 | 0 | — |
case-08 | pass→pass | 15,569 | 13,073 | -16% | 1 | 1 | 0% | 2,345 | 2,506 | +7% | 0 | 0 | — |
case-09 | pass→pass | 16,284 | 19,271 | +18% | 1 | 1 | 0% | 2,291 | 3,212 | +40% | 0 | 0 | — |
case-10 | pass→pass | 14,833 | 13,745 | -7% | 1 | 1 | 0% | 2,425 | 2,606 | +7% | 0 | 0 | — |
case-11 | pass→pass | 20,362 | 15,005 | -26% | 1 | 1 | 0% | 3,041 | 2,723 | -10% | 0 | 0 | — |
case-12 | pass→pass | 11,220 | 15,483 | +38% | 1 | 1 | 0% | 2,000 | 2,957 | +48% | 0 | 0 | — |
case-13 | pass→pass | 19,714 | 23,839 | +21% | 1 | 1 | 0% | 3,017 | 4,240 | +41% | 0 | 0 | — |
case-14 | fail→pass | 10,155 | 11,342 | +12% | 1 | 1 | 0% | 1,839 | 2,093 | +14% | 0 | 0 | — |
case-15 | pass→pass | 18,257 | 15,800 | -13% | 1 | 1 | 0% | 2,496 | 2,786 | +12% | 0 | 0 | — |
case-16 | pass→pass | 14,278 | 18,334 | +28% | 1 | 1 | 0% | 2,242 | 2,852 | +27% | 0 | 0 | — |
case-17 | pass→pass | 14,453 | 20,029 | +39% | 1 | 1 | 0% | 2,391 | 3,158 | +32% | 0 | 0 | — |
case-18 | pass→pass | 12,745 | 14,297 | +12% | 1 | 1 | 0% | 2,030 | 2,258 | +11% | 0 | 0 | — |
case-19 | pass→pass | 6,227 | 5,264 | -15% | 1 | 1 | 0% | 823 | 998 | +21% | 0 | 0 | — |
case-20 | pass→pass | 20,304 | 22,240 | +10% | 1 | 1 | 0% | 3,046 | 3,355 | +10% | 0 | 0 | — |
case-21 | pass→pass | 13,651 | 10,924 | -20% | 1 | 1 | 0% | 2,127 | 1,943 | -9% | 0 | 0 | — |
case-22 | pass→pass | 19,418 | 24,136 | +24% | 1 | 1 | 0% | 3,282 | 3,927 | +20% | 0 | 0 | — |
case-23 | pass→pass | 20,170 | 23,248 | +15% | 1 | 1 | 0% | 3,351 | 3,681 | +10% | 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. 23 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 23 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.