Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Analyze Taiwan's manufacturing industry structure including semiconductor, electronics, machinery, and petrochemical sectors. Use this skill when the user needs to understand Taiwan's industrial landscape, evaluate manufacturing sector opportunities, assess supply chain positioning, or contextualize Taiwan in global manufacturing — even if they say 'Taiwan manufacturing overview', 'semiconductor supply chain', 'what does Taiwan make', or 'industrial analysis of Taiwan'.
.claude/skills/asgard-ai-platform-tw-manufacturing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-08 | ✓→✓ | = Same ✓ | -6% | 0% |
IRON LAW: Taiwan Manufacturing = Global Supply Chain Chokepoint
Taiwan's manufacturing is NOT just domestic industry — it's a critical
node in global supply chains. TSMC alone produces ~90% of the world's
most advanced chips. Analyzing Taiwan manufacturing without the global
supply chain context misses the point entirely.| Sector | Global Position | Key Companies | Revenue Scale | |--------|----------------|-------------|-------------| | Semiconductor | #1 foundry (TSMC), #1 packaging (ASE) | TSMC, UMC, ASE, MediaTek | ~NT$4T+ | | Electronics/ICT | #1 laptop/server ODM | Foxconn, Quanta, Pegatron, Wistron | ~NT$10T+ | | Flat panel display | #2 (after China) | AUO, Innolux | ~NT$500B | | Machinery | Major machine tool exporter | Hiwin, Tongtai, Fair Friend | ~NT$1T | | Petrochemical | Top 10 globally | Formosa Plastics Group, CPC | ~NT$2T | | Bicycle | #1 premium bicycle maker | Giant, Merida | ~NT$100B | | Textile | Leading functional fabric | Far Eastern, Eclat | ~NT$300B |
For any Taiwan manufacturing sector:
markdown# Taiwan Manufacturing Sector Analysis: {Sector} ## Sector Overview - Global position: {ranking, market share} - Key players: {top 5 companies} - Revenue: {NT$ scale} ## Value Chain Position {Where Taiwan sits: upstream/midstream/downstream} ## Key Trends 1. {trend + impact} 2. {trend} ## Risks | Risk | Likelihood | Impact | Mitigation | |------|-----------|--------|-----------| | {risk} | H/M/L | H/M/L | {how addressed} | ## Opportunities 1. {opportunity for businesses}
references/semiconductor-chain.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 117,028 | 59,114 | -49% | 1 | 1 | 0% | 6,675 | 4,751 | -29% | 0 | 0 | — |
case-02 | fail→pass | 58,850 | 51,953 | -12% | 1 | 1 | 0% | 3,905 | 3,947 | +1% | 0 | 0 | — |
case-03 | fail→fail | 46,623 | 19,635 | -58% | 1 | 1 | 0% | 4,686 | 3,846 | -18% | 0 | 0 | — |
case-04 | fail→fail | 21,838 | 13,669 | -37% | 1 | 1 | 0% | 3,500 | 3,032 | -13% | 0 | 0 | — |
case-05 | fail→fail | 25,735 | 20,972 | -19% | 1 | 1 | 0% | 3,282 | 3,647 | +11% | 0 | 0 | — |
case-06 | fail→fail | 20,485 | 23,240 | +13% | 1 | 1 | 0% | 2,925 | 4,150 | +42% | 0 | 0 | — |
case-07 | fail→fail | 28,845 | 24,889 | -14% | 1 | 1 | 0% | 3,783 | 4,167 | +10% | 0 | 0 | — |
case-08 | pass→pass | 28,799 | 20,323 | -29% | 1 | 1 | 0% | 4,540 | 4,290 | -6% | 0 | 0 | — |
case-09 | fail→pass | 18,957 | 16,998 | -10% | 1 | 1 | 0% | 2,922 | 3,578 | +22% | 0 | 0 | — |
case-10 | fail→pass | 20,276 | 17,705 | -13% | 1 | 1 | 0% | 2,985 | 3,643 | +22% | 0 | 0 | — |
case-11 | pass→pass | 17,683 | 21,887 | +24% | 1 | 1 | 0% | 2,658 | 4,327 | +63% | 0 | 0 | — |
case-12 | fail→fail | 29,444 | 15,633 | -47% | 1 | 1 | 0% | 3,856 | 3,338 | -13% | 0 | 0 | — |
case-13 | fail→fail | 17,654 | 21,720 | +23% | 1 | 1 | 0% | 2,800 | 4,200 | +50% | 0 | 0 | — |
case-14 | fail→fail | 29,109 | 25,886 | -11% | 1 | 1 | 0% | 3,733 | 4,606 | +23% | 0 | 0 | — |
case-15 | pass→pass | 24,887 | 21,167 | -15% | 1 | 1 | 0% | 3,196 | 4,230 | +32% | 0 | 0 | — |
case-16 | fail→fail | 35,214 | 27,865 | -21% | 1 | 1 | 0% | 4,623 | 5,215 | +13% | 0 | 0 | — |
case-17 | fail→fail | 14,345 | 19,748 | +38% | 1 | 1 | 0% | 2,064 | 3,693 | +79% | 0 | 0 | — |
case-18 | fail→fail | 25,782 | 22,584 | -12% | 1 | 1 | 0% | 3,672 | 4,254 | +16% | 0 | 0 | — |
case-19 | fail→fail | 22,225 | 17,197 | -23% | 1 | 1 | 0% | 2,985 | 3,670 | +23% | 0 | 0 | — |
case-20 | fail→pass | 21,168 | 25,039 | +18% | 1 | 1 | 0% | 2,727 | 4,796 | +76% | 0 | 0 | — |
case-21 | fail→fail | 20,124 | 20,725 | +3% | 1 | 1 | 0% | 3,240 | 4,222 | +30% | 0 | 0 | — |
case-22 | pass→pass | 18,815 | 39,540 | +110% | 1 | 1 | 0% | 2,858 | 4,202 | +47% | 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 +18 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.