Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →用于银行零售客户存款流失预警，当需要监测余额下滑、资金流出与活跃度下降并输出处置建议时触发。
.claude/skills/aifinlab-bank-t164-retail-finance-deposit-churn-warning-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 93% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 29% | 0% |
对零售客户存款余额与资金流向进行预警监测，识别异常流出、趋势性下滑与潜在流失风险，输出风险分层、驱动因素、处置建议与升级条件，支持客户经理与运营团队进行挽留与跟进。
scripts/deposit_churn_warning.py：计算余额趋势、净流出比例并输出风险分层。bashpython scripts/deposit_churn_warning.py --input deposit.json --rules rules.json --output out.json
json{ "customer_id": "C1001", "balances": [ {"date": "2026-01-01", "balance": 500000}, {"date": "2026-02-01", "balance": 420000}, {"date": "2026-03-01", "balance": 350000} ], "flows": { "inflow_30d": 60000, "outflow_30d": 180000 } }
json{ "thresholds": { "balance_drop_pct": 0.2, "outflow_ratio": 0.5 } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | pass→pass | 9,264 | 13,495 | +46% | 1 | 1 | 0% | 1,211 | 2,367 | +95% | 0 | 0 | — |
case-01 | pass→pass | 14,746 | 28,573 | +94% | 1 | 1 | 0% | 2,453 | 3,377 | +38% | 0 | 0 | — |
case-02 | fail→fail | 14,436 | 16,455 | +14% | 1 | 1 | 0% | 2,257 | 3,167 | +40% | 0 | 0 | — |
case-03 | pass→pass | 22,416 | 24,551 | +10% | 1 | 1 | 0% | 3,183 | 3,945 | +24% | 0 | 0 | — |
case-04 | fail→pass | 7,144 | 3,763 | -47% | 1 | 1 | 0% | 1,206 | 1,112 | -8% | 0 | 0 | — |
case-10 | pass→pass | 16,714 | 18,580 | +11% | 1 | 1 | 0% | 2,462 | 3,203 | +30% | 0 | 0 | — |
case-05 | fail→pass | 17,918 | 6,868 | -62% | 1 | 1 | 0% | 2,653 | 2,114 | -20% | 0 | 0 | — |
case-06 | fail→fail | 14,789 | 22,623 | +53% | 1 | 1 | 0% | 2,612 | 3,387 | +30% | 0 | 0 | — |
case-07 | fail→pass | 18,559 | 20,057 | +8% | 1 | 1 | 0% | 2,903 | 3,282 | +13% | 0 | 0 | — |
case-08 | pass→pass | 13,556 | 17,645 | +30% | 1 | 1 | 0% | 1,876 | 3,487 | +86% | 0 | 0 | — |
case-09 | pass→pass | 20,464 | 15,471 | -24% | 1 | 1 | 0% | 3,175 | 3,350 | +6% | 0 | 0 | — |
case-11 | fail→pass | 12,475 | 23,009 | +84% | 1 | 1 | 0% | 1,800 | 3,479 | +93% | 0 | 0 | — |
case-12 | fail→pass | 17,389 | 20,282 | +17% | 1 | 1 | 0% | 2,436 | 3,151 | +29% | 0 | 0 | — |
case-13 | pass→pass | 16,164 | 18,788 | +16% | 1 | 1 | 0% | 2,073 | 3,457 | +67% | 0 | 0 | — |
case-14 | fail→pass | 12,727 | 17,474 | +37% | 1 | 1 | 0% | 1,862 | 2,977 | +60% | 0 | 0 | — |
case-15 | fail→pass | 12,995 | 12,309 | -5% | 1 | 1 | 0% | 1,775 | 2,721 | +53% | 0 | 0 | — |
case-16 | pass→pass | 19,392 | 15,464 | -20% | 1 | 1 | 0% | 2,512 | 3,120 | +24% | 0 | 0 | — |
case-18 | pass→pass | 17,874 | 16,868 | -6% | 1 | 1 | 0% | 2,355 | 3,326 | +41% | 0 | 0 | — |
case-19 | fail→pass | 21,015 | 14,298 | -32% | 1 | 1 | 0% | 2,331 | 2,817 | +21% | 0 | 0 | — |
case-20 | pass→pass | 7,108 | 12,675 | +78% | 1 | 1 | 0% | 1,135 | 2,984 | +163% | 0 | 0 | — |
case-21 | pass→pass | 15,776 | 16,786 | +6% | 1 | 1 | 0% | 2,550 | 2,904 | +14% | 0 | 0 | — |
case-22 | pass→pass | 19,587 | 20,330 | +4% | 1 | 1 | 0% | 2,608 | 3,717 | +43% | 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 +36 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.