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，围绕贷前扫描做初筛、尽调准备或审批前风险梳理时使用本技能，输出可直接用于风控与预警团队的结构化判断、待补资料清单和下一步处置建议。
.claude/skills/aifinlab-bank-t196-risk-management-pre-loan-scan-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 4% | 0% |
本技能面向银行风控与预警团队，在贷前阶段把规则命中、预警信号、经营趋势和资料一致性问题转成“可执行的风险视图”。它不是授信审批结论工具，而是为后续尽调、准入、审查提供统一的风险清单与证据链。
当需要把多条扫描信号转换为统一口径的风险摘要时，使用 scripts/pre_loan_scan_report.py：
report.json（结构化风险列表）report.md（可直接复制进报告的草稿）| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 23,442 | 31,469 | +34% | 1 | 1 | 0% | 3,347 | 4,637 | +39% | 0 | 0 | — |
case-09 | pass→pass | 21,512 | 25,322 | +18% | 1 | 1 | 0% | 2,734 | 4,284 | +57% | 0 | 0 | — |
case-10 | fail→pass | 20,015 | 21,309 | +6% | 1 | 1 | 0% | 2,894 | 4,187 | +45% | 0 | 0 | — |
case-11 | pass→pass | 23,318 | 19,563 | -16% | 1 | 1 | 0% | 3,350 | 4,146 | +24% | 0 | 0 | — |
case-12 | pass→pass | 22,303 | 20,240 | -9% | 1 | 1 | 0% | 2,806 | 3,947 | +41% | 0 | 0 | — |
case-07 | fail→pass | 15,722 | 5,399 | -66% | 1 | 1 | 0% | 2,152 | 1,641 | -24% | 0 | 0 | — |
case-02 | fail→pass | 25,017 | 27,973 | +12% | 1 | 1 | 0% | 3,184 | 5,088 | +60% | 0 | 0 | — |
case-03 | fail→pass | 16,915 | 12,695 | -25% | 1 | 1 | 0% | 2,616 | 2,725 | +4% | 0 | 0 | — |
case-04 | fail→pass | 19,631 | 15,341 | -22% | 1 | 1 | 0% | 2,405 | 2,937 | +22% | 0 | 0 | — |
case-05 | fail→pass | 32,639 | 25,862 | -21% | 1 | 1 | 0% | 4,276 | 4,712 | +10% | 0 | 0 | — |
case-06 | pass→pass | 25,648 | 24,347 | -5% | 1 | 1 | 0% | 3,094 | 3,893 | +26% | 0 | 0 | — |
case-08 | fail→pass | 23,385 | 27,331 | +17% | 1 | 1 | 0% | 3,317 | 4,996 | +51% | 0 | 0 | — |
case-13 | pass→pass | 27,692 | 24,264 | -12% | 1 | 1 | 0% | 3,393 | 4,240 | +25% | 0 | 0 | — |
case-14 | pass→pass | 18,259 | 22,768 | +25% | 1 | 1 | 0% | 2,335 | 4,460 | +91% | 0 | 0 | — |
case-15 | pass→pass | 17,386 | 18,195 | +5% | 1 | 1 | 0% | 2,611 | 3,620 | +39% | 0 | 0 | — |
case-16 | fail→pass | 14,078 | 18,638 | +32% | 1 | 1 | 0% | 2,342 | 3,822 | +63% | 0 | 0 | — |
case-17 | pass→pass | 18,333 | 27,115 | +48% | 1 | 1 | 0% | 3,028 | 4,562 | +51% | 0 | 0 | — |
case-18 | fail→pass | 23,982 | 38,257 | +60% | 1 | 1 | 0% | 3,241 | 5,002 | +54% | 0 | 0 | — |
case-19 | pass→pass | 22,332 | 22,796 | +2% | 1 | 1 | 0% | 2,839 | 4,379 | +54% | 0 | 0 | — |
case-20 | pass→pass | 23,278 | 28,569 | +23% | 1 | 1 | 0% | 3,329 | 4,854 | +46% | 0 | 0 | — |
case-21 | fail→pass | 19,804 | 26,262 | +33% | 1 | 1 | 0% | 2,777 | 4,364 | +57% | 0 | 0 | — |
case-22 | fail→pass | 23,541 | 21,494 | -9% | 1 | 1 | 0% | 2,905 | 4,183 | +44% | 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 +55 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.