Loading skill
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，对平台企业授信做财政依赖度、市场化现金流、政府支持边界、隐性债务压力和持续融资能力判断时使用本技能。适合服务政府融资平台相关客户经理和审查人员。
.claude/skills/aifinlab-bank-t129-corporate-finance-credit-platform-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-22 | ✓→✗ | ▼ Worse | 81% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 51% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 59% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 157% | 0% |
这个 skill 重点处理平台企业授信。判断核心不是“有没有国资背景”，而是平台自身经营现金流是否独立、财政支持边界是否清楚、隐性债务压力是否可控，以及融资安排是否过度依赖续借和政府兜底口径。
详细字段见 input-schema.md 和 output-schema.md。
scripts/run_skill.py：输出平台企业授信判断包shared/corporate_credit_skill_engine.py：共享分析引擎，内含平台专项规则| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 29,711 | 63,909 | +115% | 1 | 1 | 0% | 3,855 | 5,803 | +51% | 0 | 0 | — |
case-02 | pass→pass | 20,810 | 22,250 | +7% | 1 | 1 | 0% | 2,900 | 4,622 | +59% | 0 | 0 | — |
case-03 | pass→pass | 11,784 | 29,135 | +147% | 1 | 1 | 0% | 1,737 | 4,471 | +157% | 0 | 0 | — |
case-04 | fail→fail | 19,486 | 19,562 | +0% | 1 | 1 | 0% | 2,384 | 3,919 | +64% | 0 | 0 | — |
case-05 | pass→pass | 22,530 | 24,119 | +7% | 1 | 1 | 0% | 3,197 | 4,802 | +50% | 0 | 0 | — |
case-06 | pass→pass | 20,641 | 26,231 | +27% | 1 | 1 | 0% | 2,718 | 4,551 | +67% | 0 | 0 | — |
case-07 | pass→pass | 22,265 | 22,724 | +2% | 1 | 1 | 0% | 3,192 | 4,714 | +48% | 0 | 0 | — |
case-08 | pass→pass | 13,590 | 17,486 | +29% | 1 | 1 | 0% | 1,627 | 3,878 | +138% | 0 | 0 | — |
case-09 | pass→pass | 18,475 | 22,449 | +22% | 1 | 1 | 0% | 2,527 | 4,701 | +86% | 0 | 0 | — |
case-10 | pass→pass | 30,045 | 22,472 | -25% | 1 | 1 | 0% | 3,662 | 4,710 | +29% | 0 | 0 | — |
case-11 | pass→pass | 18,474 | 21,008 | +14% | 1 | 1 | 0% | 2,499 | 4,494 | +80% | 0 | 0 | — |
case-12 | pass→pass | 23,346 | 21,424 | -8% | 1 | 1 | 0% | 3,197 | 4,380 | +37% | 0 | 0 | — |
case-13 | fail→fail | 20,484 | 22,189 | +8% | 1 | 1 | 0% | 2,579 | 4,291 | +66% | 0 | 0 | — |
case-14 | pass→pass | 22,905 | 21,337 | -7% | 1 | 1 | 0% | 2,914 | 4,000 | +37% | 0 | 0 | — |
case-15 | pass→pass | 16,981 | 20,760 | +22% | 1 | 1 | 0% | 2,549 | 3,883 | +52% | 0 | 0 | — |
case-16 | pass→pass | 22,721 | 29,093 | +28% | 1 | 1 | 0% | 2,868 | 4,843 | +69% | 0 | 0 | — |
case-17 | pass→pass | 18,742 | 24,268 | +29% | 1 | 1 | 0% | 2,340 | 4,507 | +93% | 0 | 0 | — |
case-18 | pass→pass | 16,591 | 18,683 | +13% | 1 | 1 | 0% | 2,409 | 4,134 | +72% | 0 | 0 | — |
case-19 | fail→pass | 19,796 | 21,020 | +6% | 1 | 1 | 0% | 2,688 | 4,420 | +64% | 0 | 0 | — |
case-20 | pass→pass | 23,010 | 21,899 | -5% | 1 | 1 | 0% | 3,000 | 4,819 | +61% | 0 | 0 | — |
case-21 | pass→pass | 27,315 | 25,601 | -6% | 1 | 1 | 0% | 4,091 | 5,282 | +29% | 0 | 0 | — |
case-22 | pass→fail | 23,985 | 31,638 | +32% | 1 | 1 | 0% | 3,256 | 5,890 | +81% | 0 | 0 | — |
case-23 | pass→pass | 40,121 | 24,936 | -38% | 1 | 1 | 0% | 3,149 | 5,394 | +71% | 0 | 0 | — |
case-24 | pass→pass | 30,658 | 36,617 | +19% | 1 | 1 | 0% | 4,812 | 6,718 | +40% | 0 | 0 | — |
case-25 | pass→pass | 33,535 | 31,868 | -5% | 1 | 1 | 0% | 6,488 | 7,630 | +18% | 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 0 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.