Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Issue Taiwan e-invoices via UniversalEC (汎宇電商) using mcp-universalec-e-invoice (27 tools). Use when a merchant uses UniversalEC as their 加值服務中心 instead of ezPay, or when migrating between centers. Do NOT use for ezPay flows (`tw-ecom-invoice-ezpay`) or MOF direct integration (`tw-einvoice-guide`). STATUS: SKELETON — body pending.
.claude/skills/asgard-ai-platform-tw-ecom-invoice-universalec/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 23% | 0% |
> STATUS: SKELETON — body pending.
tw-ecom-invoice-ezpaytw-einvoice-guideTODO: UniversalEC market position, tool-set breadth.
TODO: UniversalEC vs ezPay trade-offs.
TODO: the 27 tool categories, common flows.
TODO.
TODO.
TODO.
tw-ecom-invoice-ezpay (parallel skill)tw-einvoice-guide_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 | 26,075 | 22,154 | -15% | 1 | 1 | 0% | 4,450 | 3,832 | -14% | 0 | 0 | — |
case-02 | fail→fail | 17,580 | 16,798 | -4% | 1 | 1 | 0% | 2,444 | 3,033 | +24% | 0 | 0 | — |
case-03 | pass→pass | 16,638 | 14,815 | -11% | 1 | 1 | 0% | 2,575 | 2,608 | +1% | 0 | 0 | — |
case-04 | fail→pass | 15,194 | 16,021 | +5% | 1 | 1 | 0% | 2,660 | 2,688 | +1% | 0 | 0 | — |
case-05 | pass→pass | 12,253 | 13,014 | +6% | 1 | 1 | 0% | 2,131 | 2,533 | +19% | 0 | 0 | — |
case-06 | pass→fail | 9,962 | 10,028 | +1% | 1 | 1 | 0% | 1,636 | 2,016 | +23% | 0 | 0 | — |
case-07 | pass→pass | 11,898 | 11,356 | -5% | 1 | 1 | 0% | 2,202 | 2,150 | -2% | 0 | 0 | — |
case-08 | pass→pass | 11,536 | 8,887 | -23% | 1 | 1 | 0% | 2,077 | 1,847 | -11% | 0 | 0 | — |
case-09 | pass→pass | 6,934 | 8,358 | +21% | 1 | 1 | 0% | 1,222 | 1,830 | +50% | 0 | 0 | — |
case-10 | pass→pass | 7,155 | 7,610 | +6% | 1 | 1 | 0% | 1,182 | 1,522 | +29% | 0 | 0 | — |
case-20 | pass→pass | 15,669 | 12,187 | -22% | 1 | 1 | 0% | 2,231 | 2,442 | +9% | 0 | 0 | — |
case-11 | pass→pass | 12,978 | 17,845 | +38% | 1 | 1 | 0% | 2,495 | 3,148 | +26% | 0 | 0 | — |
case-12 | pass→pass | 7,634 | 6,900 | -10% | 1 | 1 | 0% | 1,102 | 1,379 | +25% | 0 | 0 | — |
case-13 | fail→pass | 13,087 | 10,410 | -20% | 1 | 1 | 0% | 1,870 | 1,774 | -5% | 0 | 0 | — |
case-14 | pass→pass | 9,110 | 12,555 | +38% | 1 | 1 | 0% | 1,515 | 2,034 | +34% | 0 | 0 | — |
case-15 | pass→pass | 14,098 | 16,502 | +17% | 1 | 1 | 0% | 1,884 | 2,769 | +47% | 0 | 0 | — |
case-16 | fail→pass | 14,219 | 14,611 | +3% | 1 | 1 | 0% | 2,172 | 2,633 | +21% | 0 | 0 | — |
case-17 | pass→pass | 29,263 | 10,470 | -64% | 1 | 1 | 0% | 1,634 | 2,034 | +24% | 0 | 0 | — |
case-18 | fail→pass | 15,105 | 15,011 | -1% | 1 | 1 | 0% | 1,978 | 2,480 | +25% | 0 | 0 | — |
case-19 | pass→pass | 13,854 | 13,016 | -6% | 1 | 1 | 0% | 2,864 | 2,655 | -7% | 0 | 0 | — |
case-21 | pass→pass | 18,253 | 21,113 | +16% | 1 | 1 | 0% | 2,572 | 3,232 | +26% | 0 | 0 | — |
case-22 | pass→pass | 8,307 | 7,908 | -5% | 1 | 1 | 0% | 1,336 | 1,604 | +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 +14 percentage points is the difference between those two pass rates over the 22 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.