Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行风险管理场景下，对押品价值、覆盖度与处置可行性进行持续监测、预警识别和处置建议输出时使用本技能。适合生成押品风险摘要、优先级判断、处置动作清单与升级路径。
.claude/skills/aifinlab-bank-t201-risk-management-collateral-monitor-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 38% | 0% |
本技能面向银行风险管理场景，专注于押品价值、折扣率、覆盖度和处置可行性的监测与异常识别。目标是将规则命中、估值变化、市场事件和处置记录串成可执行的风险视图，便于风控、授信管理和预警运营团队直接采取行动。
scripts/collateral_monitor.py：用于押品价值与覆盖度的异常识别、分级与报告摘要生成| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 25,635 | 37,267 | +45% | 1 | 1 | 0% | 4,030 | 4,696 | +17% | 0 | 0 | — |
case-02 | fail→fail | 27,921 | 28,551 | +2% | 1 | 1 | 0% | 3,911 | 4,804 | +23% | 0 | 0 | — |
case-03 | fail→pass | 29,355 | 23,320 | -21% | 1 | 1 | 0% | 3,967 | 4,495 | +13% | 0 | 0 | — |
case-04 | pass→pass | 21,486 | 23,344 | +9% | 1 | 1 | 0% | 3,087 | 4,250 | +38% | 0 | 0 | — |
case-05 | pass→pass | 23,959 | 16,284 | -32% | 1 | 1 | 0% | 3,595 | 3,329 | -7% | 0 | 0 | — |
case-06 | pass→pass | 24,262 | 20,596 | -15% | 1 | 1 | 0% | 3,343 | 3,399 | +2% | 0 | 0 | — |
case-07 | pass→pass | 22,215 | 30,392 | +37% | 1 | 1 | 0% | 3,414 | 4,878 | +43% | 0 | 0 | — |
case-08 | fail→pass | 26,808 | 22,560 | -16% | 1 | 1 | 0% | 3,368 | 3,882 | +15% | 0 | 0 | — |
case-09 | pass→pass | 33,914 | 27,490 | -19% | 1 | 1 | 0% | 3,935 | 4,802 | +22% | 0 | 0 | — |
case-10 | fail→fail | 22,526 | 27,194 | +21% | 1 | 1 | 0% | 3,133 | 4,340 | +39% | 0 | 0 | — |
case-11 | pass→pass | 27,770 | 26,117 | -6% | 1 | 1 | 0% | 3,366 | 4,371 | +30% | 0 | 0 | — |
case-12 | fail→pass | 23,738 | 21,713 | -9% | 1 | 1 | 0% | 3,038 | 3,577 | +18% | 0 | 0 | — |
case-13 | fail→pass | 23,061 | 24,676 | +7% | 1 | 1 | 0% | 2,873 | 3,978 | +38% | 0 | 0 | — |
case-14 | pass→pass | 25,991 | 24,895 | -4% | 1 | 1 | 0% | 3,319 | 4,119 | +24% | 0 | 0 | — |
case-15 | fail→pass | 27,091 | 25,224 | -7% | 1 | 1 | 0% | 2,864 | 3,842 | +34% | 0 | 0 | — |
case-16 | fail→fail | 22,960 | 26,625 | +16% | 1 | 1 | 0% | 3,108 | 3,924 | +26% | 0 | 0 | — |
case-17 | fail→fail | 27,152 | 27,107 | -0% | 1 | 1 | 0% | 3,851 | 4,529 | +18% | 0 | 0 | — |
case-18 | fail→fail | 21,727 | 27,639 | +27% | 1 | 1 | 0% | 2,788 | 4,183 | +50% | 0 | 0 | — |
case-19 | pass→pass | 20,436 | 20,948 | +3% | 1 | 1 | 0% | 3,398 | 3,820 | +12% | 0 | 0 | — |
case-20 | fail→pass | 29,164 | 32,534 | +12% | 1 | 1 | 0% | 3,662 | 4,860 | +33% | 0 | 0 | — |
case-21 | pass→pass | 22,090 | 22,620 | +2% | 1 | 1 | 0% | 2,891 | 3,774 | +31% | 0 | 0 | — |
case-22 | pass→pass | 24,431 | 25,340 | +4% | 1 | 1 | 0% | 3,286 | 3,935 | +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 +32 percentage points is the difference between those two pass rates over the 22 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.