Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Integrate and operate 91APP in Taiwan e-commerce context via mcp-91app. Use when syncing orders/products/members on 91APP, running OMO (online-merge-offline) flows, integrating 91APP's app-first storefronts, or comparing 91APP vs Shopline/Shopify for mid-market DTC. Do NOT use for Shopline-specific integration (see tw-ecom-dtc-shopline). STATUS: SKELETON — body pending.
.claude/skills/asgard-ai-platform-tw-ecom-dtc-91app/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 34% | 0% |
> STATUS: SKELETON — body pending.
tw-ecom-channel-strategyTODO: 91APP app-first positioning, OMO model, member-centric architecture.
TODO: which mcp-91app tool for which task.
TODO: order sync, member sync, promotion setup, inventory sync flows.
TODO: 5-6 pitfalls (candidates: OMO state drift, promotion stacking rules, member merge across web/app, quota limits).
TODO: one non-obvious constraint.
TODO.
tw-ecom-channel-strategytw-ecom-dtc-shopline (methodology template)_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 | fail→fail | 21,648 | 54,576 | +152% | 1 | 1 | 0% | 2,892 | 3,972 | +37% | 0 | 0 | — |
case-02 | fail→fail | 13,046 | 23,435 | +80% | 1 | 1 | 0% | 2,687 | 5,396 | +101% | 0 | 0 | — |
case-03 | pass→pass | 19,399 | 23,661 | +22% | 1 | 1 | 0% | 3,122 | 3,558 | +14% | 0 | 0 | — |
case-04 | fail→pass | 20,257 | 22,119 | +9% | 1 | 1 | 0% | 3,089 | 3,531 | +14% | 0 | 0 | — |
case-05 | pass→pass | 11,056 | 13,521 | +22% | 1 | 1 | 0% | 1,853 | 2,578 | +39% | 0 | 0 | — |
case-06 | pass→pass | 18,844 | 20,023 | +6% | 1 | 1 | 0% | 2,566 | 3,110 | +21% | 0 | 0 | — |
case-07 | fail→pass | 18,640 | 14,835 | -20% | 1 | 1 | 0% | 2,936 | 2,982 | +2% | 0 | 0 | — |
case-08 | pass→pass | 16,350 | 15,066 | -8% | 1 | 1 | 0% | 2,384 | 2,550 | +7% | 0 | 0 | — |
case-09 | fail→pass | 17,472 | 19,408 | +11% | 1 | 1 | 0% | 2,667 | 2,913 | +9% | 0 | 0 | — |
case-10 | fail→pass | 19,837 | 15,359 | -23% | 1 | 1 | 0% | 2,903 | 2,513 | -13% | 0 | 0 | — |
case-11 | pass→pass | 24,138 | 29,852 | +24% | 1 | 1 | 0% | 3,846 | 5,107 | +33% | 0 | 0 | — |
case-12 | pass→pass | 14,833 | 16,006 | +8% | 1 | 1 | 0% | 2,417 | 2,592 | +7% | 0 | 0 | — |
case-13 | pass→pass | 18,462 | 20,246 | +10% | 1 | 1 | 0% | 2,795 | 3,552 | +27% | 0 | 0 | — |
case-14 | pass→pass | 23,614 | 21,846 | -7% | 1 | 1 | 0% | 3,465 | 3,906 | +13% | 0 | 0 | — |
case-15 | fail→fail | 17,348 | 12,982 | -25% | 1 | 1 | 0% | 2,413 | 2,261 | -6% | 0 | 0 | — |
case-16 | pass→pass | 13,723 | 13,754 | +0% | 1 | 1 | 0% | 2,093 | 2,299 | +10% | 0 | 0 | — |
case-17 | pass→pass | 21,698 | 22,066 | +2% | 1 | 1 | 0% | 3,165 | 3,547 | +12% | 0 | 0 | — |
case-18 | fail→pass | 17,124 | 19,922 | +16% | 1 | 1 | 0% | 2,417 | 3,229 | +34% | 0 | 0 | — |
case-19 | pass→pass | 17,603 | 15,908 | -10% | 1 | 1 | 0% | 2,729 | 2,814 | +3% | 0 | 0 | — |
case-20 | pass→pass | 18,392 | 12,688 | -31% | 1 | 1 | 0% | 3,199 | 2,495 | -22% | 0 | 0 | — |
case-25 | pass→pass | 18,985 | 20,921 | +10% | 1 | 1 | 0% | 2,797 | 3,063 | +10% | 0 | 0 | — |
case-21 | pass→pass | 18,467 | 17,858 | -3% | 1 | 1 | 0% | 2,818 | 3,091 | +10% | 0 | 0 | — |
case-22 | pass→pass | 24,314 | 23,365 | -4% | 1 | 1 | 0% | 3,503 | 4,095 | +17% | 0 | 0 | — |
case-23 | pass→pass | 15,486 | 16,977 | +10% | 1 | 1 | 0% | 2,314 | 2,965 | +28% | 0 | 0 | — |
case-24 | fail→pass | 17,516 | 41,643 | +138% | 1 | 1 | 0% | 2,469 | 3,414 | +38% | 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. 25 cases were attempted. The headline lift of +24 percentage points is the difference between those two pass rates over the 25 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.