Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when an advisor needs to match wealth-management products to an elderly client profile, compare main and backup options, and deliver suitability notes with senior-friendly communication points and risk disclosures.
.claude/skills/aifinlab-bank-t181-wealth-management-product-suitability-senior-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-20 | ✗→✓ | ▲ Improved | 70% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-18 | ✓→✗ | ▼ Worse | 29% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 38% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 20% | 0% |
这个 skill 面向银行财富管理的老年客户适配场景，强调在适当性匹配的基础上，把目标、期限、流动性与风险边界讲清楚、讲明白，并输出可直接用于对客沟通与合规留痕的适配结论。
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 | 25,139 | 43,841 | +74% | 1 | 1 | 0% | 3,545 | 4,398 | +24% | 0 | 0 | — |
case-02 | fail→fail | 23,731 | 25,808 | +9% | 1 | 1 | 0% | 3,093 | 4,799 | +55% | 0 | 0 | — |
case-03 | pass→pass | 25,610 | 27,539 | +8% | 1 | 1 | 0% | 3,169 | 4,373 | +38% | 0 | 0 | — |
case-04 | pass→pass | 23,145 | 25,323 | +9% | 1 | 1 | 0% | 3,476 | 4,166 | +20% | 0 | 0 | — |
case-05 | fail→fail | 27,620 | 23,573 | -15% | 1 | 1 | 0% | 3,393 | 4,183 | +23% | 0 | 0 | — |
case-06 | pass→pass | 18,874 | 22,191 | +18% | 1 | 1 | 0% | 2,689 | 3,943 | +47% | 0 | 0 | — |
case-07 | pass→pass | 14,138 | 20,350 | +44% | 1 | 1 | 0% | 1,919 | 4,250 | +121% | 0 | 0 | — |
case-08 | pass→pass | 18,323 | 23,804 | +30% | 1 | 1 | 0% | 2,939 | 3,912 | +33% | 0 | 0 | — |
case-09 | pass→pass | 20,954 | 28,130 | +34% | 1 | 1 | 0% | 2,986 | 4,490 | +50% | 0 | 0 | — |
case-10 | pass→pass | 19,967 | 19,476 | -2% | 1 | 1 | 0% | 3,388 | 3,991 | +18% | 0 | 0 | — |
case-11 | pass→pass | 23,842 | 30,938 | +30% | 1 | 1 | 0% | 2,980 | 5,153 | +73% | 0 | 0 | — |
case-12 | pass→pass | 17,891 | 23,941 | +34% | 1 | 1 | 0% | 2,992 | 4,490 | +50% | 0 | 0 | — |
case-13 | fail→fail | 19,192 | 29,448 | +53% | 1 | 1 | 0% | 2,742 | 4,594 | +68% | 0 | 0 | — |
case-14 | pass→pass | 21,154 | 24,559 | +16% | 1 | 1 | 0% | 3,285 | 4,164 | +27% | 0 | 0 | — |
case-15 | fail→fail | 20,864 | 31,017 | +49% | 1 | 1 | 0% | 3,221 | 5,276 | +64% | 0 | 0 | — |
case-16 | pass→pass | 15,682 | 17,313 | +10% | 1 | 1 | 0% | 2,345 | 3,424 | +46% | 0 | 0 | — |
case-17 | pass→pass | 20,292 | 22,110 | +9% | 1 | 1 | 0% | 2,659 | 3,944 | +48% | 0 | 0 | — |
case-18 | pass→fail | 29,089 | 30,815 | +6% | 1 | 1 | 0% | 3,569 | 4,593 | +29% | 0 | 0 | — |
case-19 | pass→pass | 23,866 | 24,346 | +2% | 1 | 1 | 0% | 3,051 | 4,445 | +46% | 0 | 0 | — |
case-20 | fail→pass | 18,585 | 31,767 | +71% | 1 | 1 | 0% | 2,839 | 4,819 | +70% | 0 | 0 | — |
case-21 | fail→pass | 25,471 | 31,803 | +25% | 1 | 1 | 0% | 3,285 | 4,551 | +39% | 0 | 0 | — |
case-22 | fail→fail | 24,228 | 29,159 | +20% | 1 | 1 | 0% | 3,162 | 4,438 | +40% | 0 | 0 | — |
case-23 | fail→fail | 24,344 | 30,576 | +26% | 1 | 1 | 0% | 3,508 | 4,574 | +30% | 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. 23 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.