Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当需要在银行风险管理场景下对催收对象做分层排序、优先级判断和触达动作编排时使用本技能，适用于输出可解释的优先级、行动清单与风险边界。
.claude/skills/aifinlab-bank-t220-risk-management-collection-priority-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-23 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 38% | 0% |
本技能用于把逾期、还款行为、资产价值、触达效果等多维信息整理成可执行的催收优先级排序结果，帮助风控与催收团队在资源有限时形成清晰分层、触达节奏与复盘指标。
scripts/collection_priority_ranker.py| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 29,444 | 30,900 | +5% | 1 | 1 | 0% | 4,273 | 4,672 | +9% | 0 | 0 | — |
case-02 | fail→fail | 35,054 | 30,457 | -13% | 1 | 1 | 0% | 5,147 | 5,497 | +7% | 0 | 0 | — |
case-03 | fail→fail | 31,089 | 32,824 | +6% | 1 | 1 | 0% | 4,768 | 4,829 | +1% | 0 | 0 | — |
case-04 | fail→fail | 19,198 | 25,278 | +32% | 1 | 1 | 0% | 3,104 | 4,088 | +32% | 0 | 0 | — |
case-05 | fail→fail | 26,220 | 31,483 | +20% | 1 | 1 | 0% | 4,175 | 5,609 | +34% | 0 | 0 | — |
case-06 | fail→fail | 36,896 | 32,155 | -13% | 1 | 1 | 0% | 5,898 | 6,702 | +14% | 0 | 0 | — |
case-07 | fail→fail | 22,727 | 23,332 | +3% | 1 | 1 | 0% | 3,369 | 4,284 | +27% | 0 | 0 | — |
case-08 | fail→fail | 23,725 | 23,865 | +1% | 1 | 1 | 0% | 3,225 | 4,231 | +31% | 0 | 0 | — |
case-09 | pass→pass | 18,883 | 18,415 | -2% | 1 | 1 | 0% | 2,444 | 3,361 | +38% | 0 | 0 | — |
case-10 | fail→fail | 30,823 | 31,678 | +3% | 1 | 1 | 0% | 4,279 | 4,921 | +15% | 0 | 0 | — |
case-11 | fail→pass | 28,304 | 30,548 | +8% | 1 | 1 | 0% | 3,656 | 4,657 | +27% | 0 | 0 | — |
case-12 | fail→pass | 22,959 | 28,741 | +25% | 1 | 1 | 0% | 3,419 | 4,318 | +26% | 0 | 0 | — |
case-13 | pass→pass | 30,039 | 26,382 | -12% | 1 | 1 | 0% | 3,854 | 4,622 | +20% | 0 | 0 | — |
case-14 | fail→fail | 26,751 | 23,801 | -11% | 1 | 1 | 0% | 3,199 | 3,849 | +20% | 0 | 0 | — |
case-15 | pass→pass | 22,918 | 19,489 | -15% | 1 | 1 | 0% | 2,866 | 3,319 | +16% | 0 | 0 | — |
case-16 | fail→fail | 28,240 | 25,075 | -11% | 1 | 1 | 0% | 3,468 | 4,074 | +17% | 0 | 0 | — |
case-17 | fail→fail | 23,063 | 26,810 | +16% | 1 | 1 | 0% | 3,750 | 4,218 | +12% | 0 | 0 | — |
case-18 | pass→pass | 34,767 | 32,936 | -5% | 1 | 1 | 0% | 4,044 | 5,305 | +31% | 0 | 0 | — |
case-19 | fail→fail | 24,719 | 54,154 | +119% | 1 | 1 | 0% | 3,181 | 4,945 | +55% | 0 | 0 | — |
case-20 | fail→fail | 25,126 | 30,316 | +21% | 1 | 1 | 0% | 3,247 | 4,437 | +37% | 0 | 0 | — |
case-21 | fail→pass | 28,587 | 27,421 | -4% | 1 | 1 | 0% | 3,680 | 4,678 | +27% | 0 | 0 | — |
case-22 | fail→fail | 23,768 | 32,718 | +38% | 1 | 1 | 0% | 3,290 | 5,152 | +57% | 0 | 0 | — |
case-23 | fail→pass | 28,673 | 28,101 | -2% | 1 | 1 | 0% | 3,676 | 4,335 | +18% | 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. 23 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 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.