Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行对公金融场景下,对制造业企业授信做订单质量、产能利用率、库存与原材料波动、设备投入逻辑和还款来源判断时使用本技能。适合服务客户经理和授信审查支持岗。
.claude/skills/aifinlab-bank-t127-corporate-finance-credit-manufacturing-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✓→✗ | ▼ Worse | 68% | 0% |
| case-16 | ✓→✗ | ▼ Worse | 51% | 0% |
| case-12 | ✓→✓ | = Same ✓ | 83% | 0% |
| case-14 | ✓→✓ | = Same ✓ | 44% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 81% | 0% |
这个 skill 专门处理制造业客户。核心不是泛泛而谈财报,而是把订单、产能、设备、库存、采购和回款放到同一条生产经营链里判断授信是否成立。它适合处理“报表看着还可以,但制造链条到底顺不顺、订单是不是实、扩产值不值得做”这一类问题。
详细字段见 input-schema.md 和 output-schema.md。
scripts/run_skill.py:输出制造业授信判断包shared/corporate_credit_skill_engine.py:共享分析引擎,内含制造业专项规则| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→pass | 20,239 | 24,184 | +19% | 1 | 1 | 0% | 2,412 | 4,402 | +83% | 0 | 0 | — |
case-13 | fail→fail | 25,159 | 26,263 | +4% | 1 | 1 | 0% | 3,059 | 5,056 | +65% | 0 | 0 | — |
case-14 | pass→pass | 20,252 | 20,969 | +4% | 1 | 1 | 0% | 2,922 | 4,208 | +44% | 0 | 0 | — |
case-01 | fail→fail | 29,526 | 60,078 | +103% | 1 | 1 | 0% | 3,871 | 6,759 | +75% | 0 | 0 | — |
case-02 | fail→fail | 27,371 | 36,274 | +33% | 1 | 1 | 0% | 3,656 | 6,597 | +80% | 0 | 0 | — |
case-03 | fail→fail | 23,309 | 22,599 | -3% | 1 | 1 | 0% | 3,460 | 4,788 | +38% | 0 | 0 | — |
case-04 | pass→pass | 24,101 | 58,228 | +142% | 1 | 1 | 0% | 3,223 | 5,833 | +81% | 0 | 0 | — |
case-05 | fail→fail | 26,686 | 29,913 | +12% | 1 | 1 | 0% | 3,306 | 5,327 | +61% | 0 | 0 | — |
case-06 | pass→pass | 20,809 | 24,863 | +19% | 1 | 1 | 0% | 2,926 | 4,929 | +68% | 0 | 0 | — |
case-07 | pass→pass | 41,585 | 30,520 | -27% | 1 | 1 | 0% | 2,928 | 5,777 | +97% | 0 | 0 | — |
case-08 | pass→pass | 21,157 | 28,217 | +33% | 1 | 1 | 0% | 2,527 | 4,854 | +92% | 0 | 0 | — |
case-09 | fail→fail | 21,488 | 25,455 | +18% | 1 | 1 | 0% | 2,966 | 4,882 | +65% | 0 | 0 | — |
case-10 | pass→pass | 23,650 | 29,715 | +26% | 1 | 1 | 0% | 2,941 | 5,085 | +73% | 0 | 0 | — |
case-11 | fail→fail | 25,003 | 24,895 | -0% | 1 | 1 | 0% | 3,281 | 4,815 | +47% | 0 | 0 | — |
case-15 | pass→fail | 22,444 | 34,796 | +55% | 1 | 1 | 0% | 3,341 | 5,610 | +68% | 0 | 0 | — |
case-16 | pass→fail | 25,340 | 26,441 | +4% | 1 | 1 | 0% | 2,881 | 4,358 | +51% | 0 | 0 | — |
case-17 | pass→pass | 21,612 | 26,334 | +22% | 1 | 1 | 0% | 2,738 | 4,624 | +69% | 0 | 0 | — |
case-18 | fail→fail | 26,033 | 25,907 | -0% | 1 | 1 | 0% | 3,532 | 4,991 | +41% | 0 | 0 | — |
case-19 | fail→fail | 29,461 | 29,235 | -1% | 1 | 1 | 0% | 3,704 | 6,114 | +65% | 0 | 0 | — |
case-20 | pass→pass | 22,424 | 24,775 | +10% | 1 | 1 | 0% | 3,084 | 4,429 | +44% | 0 | 0 | — |
case-21 | pass→pass | 18,990 | 18,088 | -5% | 1 | 1 | 0% | 2,337 | 4,019 | +72% | 0 | 0 | — |
case-22 | pass→pass | 18,271 | 39,164 | +114% | 1 | 1 | 0% | 2,650 | 4,410 | +66% | 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 -100 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.