Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a private-banking advisor needs to match products to a high-net-worth or family-wealth profile, compare main and backup options, and produce suitability logic, family-balance notes, and compliant communication points.
.claude/skills/aifinlab-bank-t177-wealth-management-product-suitability-private-banking-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-23 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 57% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 56% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 74% | 0% |
这个 skill 面向私行顾问场景，强调家庭资产视角、复杂约束与多目标平衡。在完成适当性匹配的同时，输出家庭资产结构影响、税务/合规注意点与可执行沟通提纲。
scripts/run_skill.py：运行适配评估并输出 Markdown/JSONshared/wealth_management_skill_engine.py示例命令：
bashpython scripts/run_skill.py --input assets/example-input.json --format markdown
references/input-schema.mdreferences/output-schema.mdreferences/suitability-checklist.mdreferences/client-interview-outline.mdreferences/risk-disclosure-points.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 | 59,914 | 74,848 | +25% | 1 | 1 | 0% | 4,534 | 7,403 | +63% | 0 | 0 | — |
case-02 | fail→fail | 27,024 | 36,519 | +35% | 1 | 1 | 0% | 3,987 | 6,331 | +59% | 0 | 0 | — |
case-03 | pass→pass | 19,526 | 25,581 | +31% | 1 | 1 | 0% | 2,784 | 4,374 | +57% | 0 | 0 | — |
case-04 | fail→pass | 17,058 | 17,287 | +1% | 1 | 1 | 0% | 2,147 | 3,331 | +55% | 0 | 0 | — |
case-05 | pass→pass | 14,740 | 18,593 | +26% | 1 | 1 | 0% | 2,099 | 3,271 | +56% | 0 | 0 | — |
case-06 | pass→pass | 14,852 | 23,956 | +61% | 1 | 1 | 0% | 2,691 | 4,689 | +74% | 0 | 0 | — |
case-07 | pass→pass | 17,940 | 25,250 | +41% | 1 | 1 | 0% | 2,860 | 4,718 | +65% | 0 | 0 | — |
case-08 | fail→fail | 28,326 | 27,196 | -4% | 1 | 1 | 0% | 3,344 | 4,831 | +44% | 0 | 0 | — |
case-09 | pass→pass | 16,506 | 20,837 | +26% | 1 | 1 | 0% | 2,507 | 4,005 | +60% | 0 | 0 | — |
case-10 | pass→pass | 17,420 | 18,299 | +5% | 1 | 1 | 0% | 2,425 | 3,360 | +39% | 0 | 0 | — |
case-11 | pass→pass | 12,314 | 19,485 | +58% | 1 | 1 | 0% | 1,823 | 3,599 | +97% | 0 | 0 | — |
case-12 | pass→pass | 20,332 | 30,160 | +48% | 1 | 1 | 0% | 3,151 | 5,314 | +69% | 0 | 0 | — |
case-13 | fail→fail | 20,333 | 26,399 | +30% | 1 | 1 | 0% | 2,848 | 4,414 | +55% | 0 | 0 | — |
case-14 | pass→pass | 20,997 | 29,766 | +42% | 1 | 1 | 0% | 3,048 | 5,346 | +75% | 0 | 0 | — |
case-15 | pass→pass | 18,027 | 20,912 | +16% | 1 | 1 | 0% | 2,571 | 3,628 | +41% | 0 | 0 | — |
case-16 | pass→pass | 14,094 | 22,853 | +62% | 1 | 1 | 0% | 2,024 | 3,840 | +90% | 0 | 0 | — |
case-17 | fail→fail | 24,449 | 26,791 | +10% | 1 | 1 | 0% | 3,302 | 4,463 | +35% | 0 | 0 | — |
case-18 | pass→pass | 19,832 | 25,188 | +27% | 1 | 1 | 0% | 3,076 | 4,863 | +58% | 0 | 0 | — |
case-19 | pass→pass | 17,884 | 26,238 | +47% | 1 | 1 | 0% | 2,677 | 4,578 | +71% | 0 | 0 | — |
case-20 | pass→pass | 16,153 | 19,913 | +23% | 1 | 1 | 0% | 2,294 | 3,753 | +64% | 0 | 0 | — |
case-21 | pass→pass | 20,883 | 27,715 | +33% | 1 | 1 | 0% | 3,193 | 4,967 | +56% | 0 | 0 | — |
case-22 | pass→pass | 17,158 | 16,259 | -5% | 1 | 1 | 0% | 2,446 | 3,083 | +26% | 0 | 0 | — |
case-23 | fail→pass | 17,683 | 16,971 | -4% | 1 | 1 | 0% | 2,723 | 3,276 | +20% | 0 | 0 | — |
case-24 | pass→pass | 13,155 | 18,422 | +40% | 1 | 1 | 0% | 2,495 | 3,624 | +45% | 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. 24 cases were attempted. The headline lift of +8 percentage points is the difference between those two pass rates over the 24 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.