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，对跨境交易异常进行监测、分级与排查建议输出时使用本技能。适合生成异常信号摘要、优先级判断、排查清单与升级路径。
.claude/skills/aifinlab-bank-t206-risk-management-cross-border-anomaly-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 27% | 0% |
本技能面向银行风险管理场景，聚焦跨境交易的资金流向、币种结构、对手方与地缘风险异常识别。目标是将规则命中、跨境资金路径与客户画像变化转成可执行的排查清单与升级路径，支持反洗钱、合规与预警运营团队。
scripts/cross_border_anomaly_scan.py：用于跨境交易异常规则命中、分级与摘要输出| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 28,091 | 23,250 | -17% | 1 | 1 | 0% | 3,526 | 4,266 | +21% | 0 | 0 | — |
case-02 | fail→pass | 24,509 | 25,577 | +4% | 1 | 1 | 0% | 3,069 | 4,107 | +34% | 0 | 0 | — |
case-03 | fail→fail | 23,555 | 30,072 | +28% | 1 | 1 | 0% | 3,611 | 5,529 | +53% | 0 | 0 | — |
case-04 | fail→fail | 24,837 | 22,287 | -10% | 1 | 1 | 0% | 3,175 | 4,069 | +28% | 0 | 0 | — |
case-05 | fail→pass | 21,338 | 23,949 | +12% | 1 | 1 | 0% | 3,009 | 4,085 | +36% | 0 | 0 | — |
case-06 | fail→fail | 22,932 | 25,039 | +9% | 1 | 1 | 0% | 3,013 | 3,954 | +31% | 0 | 0 | — |
case-07 | pass→fail | 23,828 | 24,173 | +1% | 1 | 1 | 0% | 2,990 | 4,439 | +48% | 0 | 0 | — |
case-08 | pass→pass | 22,084 | 23,101 | +5% | 1 | 1 | 0% | 3,035 | 3,711 | +22% | 0 | 0 | — |
case-09 | fail→pass | 17,706 | 21,047 | +19% | 1 | 1 | 0% | 2,450 | 3,342 | +36% | 0 | 0 | — |
case-10 | fail→pass | 21,970 | 15,291 | -30% | 1 | 1 | 0% | 2,936 | 2,668 | -9% | 0 | 0 | — |
case-11 | fail→pass | 21,319 | 21,727 | +2% | 1 | 1 | 0% | 3,004 | 3,820 | +27% | 0 | 0 | — |
case-12 | fail→fail | 24,831 | 21,611 | -13% | 1 | 1 | 0% | 3,169 | 3,801 | +20% | 0 | 0 | — |
case-13 | pass→pass | 45,753 | 28,447 | -38% | 1 | 1 | 0% | 3,153 | 4,358 | +38% | 0 | 0 | — |
case-14 | pass→pass | 31,081 | 23,437 | -25% | 1 | 1 | 0% | 4,338 | 3,960 | -9% | 0 | 0 | — |
case-15 | fail→fail | 25,286 | 22,786 | -10% | 1 | 1 | 0% | 3,365 | 3,844 | +14% | 0 | 0 | — |
case-16 | pass→pass | 24,404 | 28,905 | +18% | 1 | 1 | 0% | 2,935 | 4,422 | +51% | 0 | 0 | — |
case-17 | fail→fail | 22,775 | 27,019 | +19% | 1 | 1 | 0% | 3,186 | 4,094 | +28% | 0 | 0 | — |
case-18 | pass→pass | 26,420 | 36,066 | +37% | 1 | 1 | 0% | 3,614 | 4,634 | +28% | 0 | 0 | — |
case-19 | fail→fail | 26,210 | 24,470 | -7% | 1 | 1 | 0% | 3,213 | 3,778 | +18% | 0 | 0 | — |
case-20 | pass→pass | 20,683 | 32,347 | +56% | 1 | 1 | 0% | 2,966 | 4,038 | +36% | 0 | 0 | — |
case-21 | fail→fail | 25,914 | 29,896 | +15% | 1 | 1 | 0% | 3,018 | 4,582 | +52% | 0 | 0 | — |
case-22 | pass→pass | 24,176 | 20,060 | -17% | 1 | 1 | 0% | 2,954 | 3,324 | +13% | 0 | 0 | — |
case-23 | pass→pass | 19,331 | 20,056 | +4% | 1 | 1 | 0% | 2,280 | 3,270 | +43% | 0 | 0 | — |
case-24 | fail→fail | 21,811 | 28,457 | +30% | 1 | 1 | 0% | 3,135 | 4,577 | +46% | 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. 24 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 24 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.