Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当需要对监控名单筛查结果进行汇总、误报排查与风险处置建议输出时使用本技能。
.claude/skills/aifinlab-bank-t218-risk-management-watchlist-screening-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -11% | 0% |
用于汇总名单筛查结果、识别重点命中对象并输出风险优先级、排查路径和升级建议，适用于监测与预警运营场景。
scripts/watchlist_screening_summary.py：生成名单筛查汇总与处置建议草稿。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | fail→pass | 24,242 | 23,285 | -4% | 1 | 1 | 0% | 3,080 | 3,608 | +17% | 0 | 0 | — |
case-11 | pass→pass | 17,260 | 18,853 | +9% | 1 | 1 | 0% | 2,604 | 3,059 | +17% | 0 | 0 | — |
case-12 | fail→fail | 12,170 | 3,669 | -70% | 1 | 1 | 0% | 1,880 | 871 | -54% | 0 | 0 | — |
case-13 | pass→pass | 22,008 | 21,639 | -2% | 1 | 1 | 0% | 2,724 | 3,005 | +10% | 0 | 0 | — |
case-20 | fail→pass | 17,119 | 14,502 | -15% | 1 | 1 | 0% | 2,458 | 2,176 | -11% | 0 | 0 | — |
case-01 | fail→fail | 25,828 | 22,700 | -12% | 1 | 1 | 0% | 3,382 | 3,654 | +8% | 0 | 0 | — |
case-02 | fail→pass | 24,927 | 23,525 | -6% | 1 | 1 | 0% | 3,835 | 3,849 | +0% | 0 | 0 | — |
case-03 | fail→pass | 27,299 | 26,657 | -2% | 1 | 1 | 0% | 3,472 | 3,751 | +8% | 0 | 0 | — |
case-04 | fail→pass | 23,741 | 20,629 | -13% | 1 | 1 | 0% | 3,054 | 2,715 | -11% | 0 | 0 | — |
case-05 | pass→pass | 24,489 | 24,625 | +1% | 1 | 1 | 0% | 3,166 | 3,944 | +25% | 0 | 0 | — |
case-06 | pass→pass | 20,368 | 17,110 | -16% | 1 | 1 | 0% | 2,309 | 2,751 | +19% | 0 | 0 | — |
case-07 | fail→pass | 12,193 | 2,463 | -80% | 1 | 1 | 0% | 1,639 | 660 | -60% | 0 | 0 | — |
case-08 | fail→pass | 23,911 | 21,120 | -12% | 1 | 1 | 0% | 2,971 | 3,311 | +11% | 0 | 0 | — |
case-09 | fail→pass | 21,788 | 18,762 | -14% | 1 | 1 | 0% | 2,979 | 2,730 | -8% | 0 | 0 | — |
case-14 | fail→pass | 22,161 | 41,394 | +87% | 1 | 1 | 0% | 2,659 | 2,238 | -16% | 0 | 0 | — |
case-15 | pass→pass | 18,203 | 18,039 | -1% | 1 | 1 | 0% | 2,582 | 2,728 | +6% | 0 | 0 | — |
case-16 | fail→fail | 20,756 | 20,671 | -0% | 1 | 1 | 0% | 2,563 | 2,923 | +14% | 0 | 0 | — |
case-17 | fail→pass | 22,639 | 18,576 | -18% | 1 | 1 | 0% | 2,913 | 3,104 | +7% | 0 | 0 | — |
case-18 | pass→pass | 17,468 | 20,767 | +19% | 1 | 1 | 0% | 2,227 | 2,696 | +21% | 0 | 0 | — |
case-19 | pass→fail | 20,135 | 20,685 | +3% | 1 | 1 | 0% | 2,764 | 3,160 | +14% | 0 | 0 | — |
case-21 | fail→fail | 13,694 | 15,807 | +15% | 1 | 1 | 0% | 1,783 | 2,586 | +45% | 0 | 0 | — |
case-22 | fail→fail | 57,867 | 30,301 | -48% | 1 | 1 | 0% | 5,221 | 5,714 | +9% | 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 +41 percentage points is the difference between those two pass rates over the 22 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.