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，围绕关联方关系图谱进行结构化梳理、风险识别与优先级判断时使用本技能。适合输出图谱摘要、风险聚合结论与排查建议。
.claude/skills/aifinlab-bank-t209-risk-management-related-party-graph-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 3% | 0% |
本技能面向银行风险管理场景，聚焦客户、账户、法人、股权与交易关系的图谱化梳理与风险识别。目标是将关联方关系、交易路径与风险标签整合为可执行的排查建议，支持风控、合规与授信审查团队。
scripts/related_party_graph_analyzer.py：用于关联关系图谱整理、风险聚合与摘要输出| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 25,672 | 32,469 | +26% | 1 | 1 | 0% | 3,627 | 4,794 | +32% | 0 | 0 | — |
case-22 | fail→fail | 25,936 | 25,416 | -2% | 1 | 1 | 0% | 3,325 | 3,754 | +13% | 0 | 0 | — |
case-03 | fail→pass | 23,439 | 17,957 | -23% | 1 | 1 | 0% | 3,334 | 3,691 | +11% | 0 | 0 | — |
case-13 | pass→pass | 18,685 | 21,320 | +14% | 1 | 1 | 0% | 2,756 | 3,909 | +42% | 0 | 0 | — |
case-01 | fail→pass | 28,943 | 31,649 | +9% | 1 | 1 | 0% | 4,339 | 4,371 | +1% | 0 | 0 | — |
case-04 | pass→pass | 26,203 | 17,082 | -35% | 1 | 1 | 0% | 3,962 | 3,930 | -1% | 0 | 0 | — |
case-05 | pass→pass | 21,660 | 24,178 | +12% | 1 | 1 | 0% | 3,735 | 4,764 | +28% | 0 | 0 | — |
case-06 | pass→pass | 20,868 | 32,444 | +55% | 1 | 1 | 0% | 3,375 | 5,162 | +53% | 0 | 0 | — |
case-07 | fail→pass | 16,728 | 3,155 | -81% | 1 | 1 | 0% | 2,444 | 1,088 | -55% | 0 | 0 | — |
case-08 | fail→pass | 17,868 | 15,433 | -14% | 1 | 1 | 0% | 2,156 | 2,615 | +21% | 0 | 0 | — |
case-09 | fail→pass | 19,849 | 18,948 | -5% | 1 | 1 | 0% | 2,906 | 2,993 | +3% | 0 | 0 | — |
case-10 | fail→pass | 22,181 | 21,113 | -5% | 1 | 1 | 0% | 3,047 | 3,687 | +21% | 0 | 0 | — |
case-11 | pass→pass | 22,161 | 19,951 | -10% | 1 | 1 | 0% | 2,994 | 3,535 | +18% | 0 | 0 | — |
case-12 | fail→fail | 16,478 | 20,462 | +24% | 1 | 1 | 0% | 2,574 | 3,289 | +28% | 0 | 0 | — |
case-14 | fail→pass | 22,193 | 18,879 | -15% | 1 | 1 | 0% | 2,737 | 3,272 | +20% | 0 | 0 | — |
case-15 | fail→pass | 22,769 | 24,343 | +7% | 1 | 1 | 0% | 3,260 | 4,227 | +30% | 0 | 0 | — |
case-16 | pass→pass | 21,696 | 28,331 | +31% | 1 | 1 | 0% | 2,825 | 3,973 | +41% | 0 | 0 | — |
case-17 | fail→pass | 16,581 | 15,985 | -4% | 1 | 1 | 0% | 1,989 | 2,909 | +46% | 0 | 0 | — |
case-18 | pass→pass | 26,775 | 55,579 | +108% | 1 | 1 | 0% | 3,550 | 4,599 | +30% | 0 | 0 | — |
case-19 | pass→pass | 27,533 | 26,287 | -5% | 1 | 1 | 0% | 3,889 | 4,308 | +11% | 0 | 0 | — |
case-20 | fail→pass | 25,721 | 25,052 | -3% | 1 | 1 | 0% | 3,461 | 4,055 | +17% | 0 | 0 | — |
case-21 | pass→pass | 21,991 | 24,662 | +12% | 1 | 1 | 0% | 2,959 | 3,808 | +29% | 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 +45 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.