Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行财富管理场景下，围绕再平衡进行适配、配置、推荐、方案设计或备选方案比较时使用本技能。适合输出主方案、备选方案、匹配逻辑、风险提示和沟通要点。
.claude/skills/aifinlab-bank-t189-wealth-management-rebalancing-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 78% | 0% |
本技能面向银行财富管理场景，核心是把客户目标、风险承受能力、期限约束、家庭资产配置和市场环境组织成可解释、可沟通、可合规落地的建议。 当前这支 skill 更偏向服务 理财经理、财富顾问和私行团队，输出时要特别注意 把客户目标、风险偏好和市场环境组织成可解释建议。
scripts/rebalance_plan.py：基于持仓与目标权重输出再平衡草案（买入/卖出方向与金额）references/rebalancing-input-schema.json：脚本输入结构示例references/rebalancing-input-schema.json）bashpython scripts/rebalance_plan.py --input input.json --output rebalance.csv --summary rebalance.txt
asset：资产名称action：买入/卖出trade_value：建议调仓金额（不含费用、税费）reason：偏离度与目标权重解释| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | pass→pass | 12,235 | 5,217 | -57% | 1 | 1 | 0% | 1,615 | 1,931 | +20% | 0 | 0 | — |
case-16 | fail→pass | 12,785 | 11,366 | -11% | 1 | 1 | 0% | 1,606 | 2,830 | +76% | 0 | 0 | — |
case-17 | pass→pass | 17,340 | 10,753 | -38% | 1 | 1 | 0% | 2,066 | 2,919 | +41% | 0 | 0 | — |
case-01 | pass→pass | 19,220 | 21,173 | +10% | 1 | 1 | 0% | 2,633 | 4,364 | +66% | 0 | 0 | — |
case-02 | pass→pass | 13,778 | 21,421 | +55% | 1 | 1 | 0% | 1,776 | 4,070 | +129% | 0 | 0 | — |
case-03 | pass→pass | 9,747 | 14,644 | +50% | 1 | 1 | 0% | 1,491 | 3,398 | +128% | 0 | 0 | — |
case-04 | fail→pass | 6,474 | 4,586 | -29% | 1 | 1 | 0% | 911 | 1,900 | +109% | 0 | 0 | — |
case-09 | fail→fail | 10,389 | 4,637 | -55% | 1 | 1 | 0% | 1,565 | 1,886 | +21% | 0 | 0 | — |
case-05 | fail→pass | 20,634 | 4,784 | -77% | 1 | 1 | 0% | 3,552 | 1,894 | -47% | 0 | 0 | — |
case-06 | fail→pass | 10,442 | 2,258 | -78% | 1 | 1 | 0% | 1,551 | 1,576 | +2% | 0 | 0 | — |
case-07 | fail→pass | 15,470 | 18,461 | +19% | 1 | 1 | 0% | 2,288 | 4,073 | +78% | 0 | 0 | — |
case-08 | fail→pass | 16,124 | 17,469 | +8% | 1 | 1 | 0% | 2,377 | 4,024 | +69% | 0 | 0 | — |
case-10 | fail→pass | 10,037 | 3,887 | -61% | 1 | 1 | 0% | 1,170 | 1,741 | +49% | 0 | 0 | — |
case-11 | fail→pass | 14,309 | 7,622 | -47% | 1 | 1 | 0% | 2,308 | 2,487 | +8% | 0 | 0 | — |
case-12 | fail→pass | 15,304 | 19,715 | +29% | 1 | 1 | 0% | 2,385 | 4,047 | +70% | 0 | 0 | — |
case-13 | pass→pass | 14,799 | 14,299 | -3% | 1 | 1 | 0% | 2,258 | 3,363 | +49% | 0 | 0 | — |
case-14 | fail→pass | 15,722 | 8,723 | -45% | 1 | 1 | 0% | 2,145 | 2,473 | +15% | 0 | 0 | — |
case-18 | pass→pass | 10,580 | 3,331 | -69% | 1 | 1 | 0% | 1,366 | 1,728 | +27% | 0 | 0 | — |
case-19 | pass→pass | 9,036 | 6,035 | -33% | 1 | 1 | 0% | 1,215 | 2,073 | +71% | 0 | 0 | — |
case-20 | fail→fail | 12,016 | 4,464 | -63% | 1 | 1 | 0% | 1,502 | 1,612 | +7% | 0 | 0 | — |
case-21 | pass→pass | 11,438 | 14,600 | +28% | 1 | 1 | 0% | 1,709 | 3,756 | +120% | 0 | 0 | — |
case-22 | fail→fail | 12,055 | 18,009 | +49% | 1 | 1 | 0% | 1,611 | 4,157 | +158% | 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.