Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a wealth manager needs to diagnose a client portfolio, explain performance drivers and concentration risks, and produce a compliant holding-diagnosis brief with follow-up actions.
.claude/skills/aifinlab-bank-t184-wealth-management-holding-diagnosis-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 37% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 66% | 0% |
这个 skill 用于银行财富管理场景的“持仓诊断”，目标是把客户画像、持仓结构、收益波动与市场环境串成可解释、可复核、可落地的诊断结论，输出可直接用于客户沟通与内部复盘的诊断简报。
参考视角：以“麦肯锡问题树式诊断 + 资产配置体检报告”的结构组织因果链和行动建议。
scripts/holding_diagnosis.py：解析持仓数据、计算结构与集中度、输出诊断包scripts/run_skill.py：命令行入口，输出 Markdown 或 JSON示例命令：
bashpython scripts/run_skill.py --input assets/example-input.json --format markdown
references/input-schema.mdreferences/output-schema.mdreferences/diagnosis-framework.mdreferences/holding-risks-checklist.mdreferences/performance-attribution-guide.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 30,327 | 66,965 | +121% | 1 | 1 | 0% | 4,657 | 6,089 | +31% | 0 | 0 | — |
case-02 | fail→pass | 29,150 | 33,427 | +15% | 1 | 1 | 0% | 3,807 | 5,848 | +54% | 0 | 0 | — |
case-03 | fail→pass | 29,501 | 27,101 | -8% | 1 | 1 | 0% | 3,629 | 4,989 | +37% | 0 | 0 | — |
case-04 | pass→pass | 16,112 | 22,963 | +43% | 1 | 1 | 0% | 2,247 | 3,818 | +70% | 0 | 0 | — |
case-05 | fail→pass | 12,079 | 10,696 | -11% | 1 | 1 | 0% | 1,865 | 2,585 | +39% | 0 | 0 | — |
case-06 | pass→pass | 17,119 | 28,493 | +66% | 1 | 1 | 0% | 2,360 | 4,645 | +97% | 0 | 0 | — |
case-07 | pass→pass | 22,766 | 18,741 | -18% | 1 | 1 | 0% | 2,994 | 4,138 | +38% | 0 | 0 | — |
case-08 | pass→pass | 29,619 | 28,000 | -5% | 1 | 1 | 0% | 3,487 | 5,341 | +53% | 0 | 0 | — |
case-09 | pass→pass | 25,283 | 41,301 | +63% | 1 | 1 | 0% | 4,313 | 7,642 | +77% | 0 | 0 | — |
case-18 | fail→fail | 24,008 | 28,451 | +19% | 1 | 1 | 0% | 3,305 | 5,098 | +54% | 0 | 0 | — |
case-19 | fail→fail | 16,442 | 19,187 | +17% | 1 | 1 | 0% | 2,364 | 3,833 | +62% | 0 | 0 | — |
case-10 | fail→pass | 16,729 | 27,572 | +65% | 1 | 1 | 0% | 2,464 | 4,569 | +85% | 0 | 0 | — |
case-11 | pass→pass | 22,243 | 24,847 | +12% | 1 | 1 | 0% | 2,966 | 4,209 | +42% | 0 | 0 | — |
case-12 | pass→pass | 26,994 | 33,706 | +25% | 1 | 1 | 0% | 3,352 | 5,025 | +50% | 0 | 0 | — |
case-13 | pass→pass | 25,137 | 32,376 | +29% | 1 | 1 | 0% | 3,265 | 6,232 | +91% | 0 | 0 | — |
case-14 | pass→pass | 19,680 | 22,566 | +15% | 1 | 1 | 0% | 2,537 | 4,358 | +72% | 0 | 0 | — |
case-15 | fail→fail | 17,433 | 22,262 | +28% | 1 | 1 | 0% | 2,479 | 3,733 | +51% | 0 | 0 | — |
case-16 | pass→pass | 23,817 | 31,729 | +33% | 1 | 1 | 0% | 3,038 | 5,373 | +77% | 0 | 0 | — |
case-17 | fail→pass | 20,282 | 22,717 | +12% | 1 | 1 | 0% | 2,799 | 4,654 | +66% | 0 | 0 | — |
case-20 | pass→pass | 24,659 | 31,482 | +28% | 1 | 1 | 0% | 3,347 | 5,965 | +78% | 0 | 0 | — |
case-21 | pass→pass | 26,090 | 30,606 | +17% | 1 | 1 | 0% | 3,418 | 4,949 | +45% | 0 | 0 | — |
case-22 | pass→pass | 20,614 | 29,107 | +41% | 1 | 1 | 0% | 2,726 | 4,638 | +70% | 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 +23 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.