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场景下对企业担保链条做风险扫描（传导路径、代偿触发点、圈层暴露、升级条件），并输出核验要点、补件清单、处置建议与升级路径时触发本技能。适合尽调/审查/贷后预警的结构化扫描。
.claude/skills/aifinlab-bank-t137-corporate-finance-task-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 220% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 68% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 108% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 77% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 46% | 0% |
担保链风险的难点不在“有没有担保”，而在“担保如何传导、何时触发、触发后谁会被拖下水”。这个 skill 用于把分散的担保信息，整理成一份可以直接用于尽调、审查与贷后监测的扫描包：
最小可用输入见 input-schema.md。
建议输出结构见 output-schema.md。
scripts/run_skill.py：运行 t137 场景，生成“担保链风险扫描包”..\shared\corporate_credit_skill_engine.py：共享引擎（统一输出结构与规则）| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | 14,233 | 6,112 | -57% | 1 | 1 | 0% | 693 | 2,221 | +220% | 0 | 0 | — |
case-07 | pass→pass | 25,043 | 39,380 | +57% | 1 | 1 | 0% | 3,602 | 6,390 | +77% | 0 | 0 | — |
case-01 | pass→pass | 30,502 | 34,158 | +12% | 1 | 1 | 0% | 4,154 | 6,047 | +46% | 0 | 0 | — |
case-02 | fail→pass | 31,702 | 39,740 | +25% | 1 | 1 | 0% | 4,677 | 7,877 | +68% | 0 | 0 | — |
case-03 | fail→fail | 58,209 | 63,279 | +9% | 1 | 1 | 0% | 4,276 | 6,623 | +55% | 0 | 0 | — |
case-04 | fail→pass | 19,743 | 29,552 | +50% | 1 | 1 | 0% | 2,933 | 6,113 | +108% | 0 | 0 | — |
case-05 | pass→pass | 18,344 | 23,843 | +30% | 1 | 1 | 0% | 2,616 | 4,349 | +66% | 0 | 0 | — |
case-08 | pass→pass | 23,719 | 27,248 | +15% | 1 | 1 | 0% | 3,327 | 5,359 | +61% | 0 | 0 | — |
case-09 | pass→pass | 20,257 | 28,595 | +41% | 1 | 1 | 0% | 2,808 | 5,817 | +107% | 0 | 0 | — |
case-10 | pass→pass | 22,050 | 22,411 | +2% | 1 | 1 | 0% | 2,808 | 4,844 | +73% | 0 | 0 | — |
case-11 | pass→pass | 18,962 | 27,535 | +45% | 1 | 1 | 0% | 2,744 | 5,050 | +84% | 0 | 0 | — |
case-12 | pass→pass | 24,178 | 28,676 | +19% | 1 | 1 | 0% | 3,204 | 5,165 | +61% | 0 | 0 | — |
case-13 | pass→pass | 22,562 | 28,257 | +25% | 1 | 1 | 0% | 2,747 | 5,415 | +97% | 0 | 0 | — |
case-14 | pass→pass | 23,808 | 27,286 | +15% | 1 | 1 | 0% | 3,597 | 5,400 | +50% | 0 | 0 | — |
case-15 | pass→pass | 21,919 | 29,528 | +35% | 1 | 1 | 0% | 2,995 | 5,863 | +96% | 0 | 0 | — |
case-16 | fail→fail | 24,623 | 26,431 | +7% | 1 | 1 | 0% | 3,049 | 5,011 | +64% | 0 | 0 | — |
case-17 | pass→pass | 22,969 | 37,574 | +64% | 1 | 1 | 0% | 2,874 | 6,344 | +121% | 0 | 0 | — |
case-18 | pass→pass | 26,165 | 38,707 | +48% | 1 | 1 | 0% | 3,757 | 6,680 | +78% | 0 | 0 | — |
case-19 | pass→pass | 25,288 | 26,661 | +5% | 1 | 1 | 0% | 3,175 | 4,892 | +54% | 0 | 0 | — |
case-20 | pass→pass | 24,634 | 28,635 | +16% | 1 | 1 | 0% | 3,381 | 5,909 | +75% | 0 | 0 | — |
case-21 | pass→pass | 24,109 | 25,979 | +8% | 1 | 1 | 0% | 2,884 | 5,228 | +81% | 0 | 0 | — |
case-22 | pass→pass | 24,093 | 37,681 | +56% | 1 | 1 | 0% | 2,966 | 7,132 | +140% | 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, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +14 percentage points is the difference between those two pass rates over the 21 comparable cases.
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.