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，对客户身份异常与资料变更进行监测、排查与处置建议输出时使用本技能。适合生成身份异常摘要、优先级判断、排查清单与升级路径。
.claude/skills/aifinlab-bank-t208-risk-management-identity-anomaly-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 30% | 0% |
本技能面向银行风险管理场景，聚焦客户身份资料、证件变更、关联信息不一致与身份核验异常识别。目标是将规则命中、身份核验结果与历史资料变更记录转成可执行的排查与处置动作，支持风控、合规与运营团队。
scripts/identity_anomaly_scan.py：用于身份异常规则命中、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 | 51,232 | 52,430 | +2% | 1 | 1 | 0% | 3,135 | 3,775 | +20% | 0 | 0 | — |
case-02 | fail→fail | 24,676 | 20,183 | -18% | 1 | 1 | 0% | 3,384 | 3,530 | +4% | 0 | 0 | — |
case-03 | pass→pass | 50,020 | 24,657 | -51% | 1 | 1 | 0% | 2,937 | 4,338 | +48% | 0 | 0 | — |
case-04 | pass→pass | 22,686 | 21,774 | -4% | 1 | 1 | 0% | 2,970 | 3,958 | +33% | 0 | 0 | — |
case-05 | fail→fail | 18,435 | 21,730 | +18% | 1 | 1 | 0% | 2,845 | 3,461 | +22% | 0 | 0 | — |
case-06 | fail→fail | 24,294 | 24,477 | +1% | 1 | 1 | 0% | 3,451 | 4,144 | +20% | 0 | 0 | — |
case-07 | fail→pass | 19,378 | 10,880 | -44% | 1 | 1 | 0% | 3,672 | 2,562 | -30% | 0 | 0 | — |
case-08 | fail→fail | 23,466 | 22,999 | -2% | 1 | 1 | 0% | 2,953 | 3,897 | +32% | 0 | 0 | — |
case-09 | fail→pass | 20,730 | 17,879 | -14% | 1 | 1 | 0% | 2,675 | 3,312 | +24% | 0 | 0 | — |
case-10 | fail→fail | 19,867 | 22,104 | +11% | 1 | 1 | 0% | 2,879 | 3,635 | +26% | 0 | 0 | — |
case-11 | pass→pass | 17,372 | 19,370 | +12% | 1 | 1 | 0% | 2,512 | 3,549 | +41% | 0 | 0 | — |
case-12 | fail→pass | 18,310 | 20,306 | +11% | 1 | 1 | 0% | 2,778 | 3,494 | +26% | 0 | 0 | — |
case-13 | fail→fail | 23,454 | 24,384 | +4% | 1 | 1 | 0% | 3,132 | 4,246 | +36% | 0 | 0 | — |
case-14 | pass→pass | 15,402 | 17,877 | +16% | 1 | 1 | 0% | 2,126 | 3,193 | +50% | 0 | 0 | — |
case-15 | pass→pass | 19,619 | 19,391 | -1% | 1 | 1 | 0% | 2,594 | 3,615 | +39% | 0 | 0 | — |
case-16 | fail→pass | 11,070 | 16,264 | +47% | 1 | 1 | 0% | 1,636 | 3,030 | +85% | 0 | 0 | — |
case-17 | fail→fail | 22,007 | 20,544 | -7% | 1 | 1 | 0% | 3,077 | 3,676 | +19% | 0 | 0 | — |
case-18 | fail→fail | 21,138 | 24,376 | +15% | 1 | 1 | 0% | 2,915 | 4,002 | +37% | 0 | 0 | — |
case-19 | pass→pass | 23,788 | 21,604 | -9% | 1 | 1 | 0% | 3,339 | 3,829 | +15% | 0 | 0 | — |
case-20 | fail→pass | 18,774 | 19,861 | +6% | 1 | 1 | 0% | 2,720 | 3,539 | +30% | 0 | 0 | — |
case-21 | fail→fail | 20,134 | 19,308 | -4% | 1 | 1 | 0% | 2,888 | 3,375 | +17% | 0 | 0 | — |
case-22 | pass→pass | 14,549 | 17,952 | +23% | 1 | 1 | 0% | 2,182 | 3,246 | +49% | 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 +23 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.