Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当需要在银行财富管理场景下输出客户陪伴周报、周度摘要与跟进计划时使用本技能，适合将多来源材料整理为结构化周度内容、重点事项与后续动作。
.claude/skills/aifinlab-bank-t192-wealth-management-customer-companion-weekly-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 29% | 0% |
面向银行财富管理场景的客户陪伴周报助手，用于将客户目标、风险偏好、持仓变化、市场事件与沟通记录等多来源材料整理成可交付的周度陪伴内容。输出以“周度节奏、重点事项、后续动作”为核心，确保合规、可执行、可追踪。
必填信息（至少满足其中三类）
可选信息
如需批量或标准化生成周报，可使用 scripts/weekly_companion_builder.py。
脚本能力：
输入 JSON 结构（示例字段）
customer_profile：客户画像risk_profile：风险等级/适当性holdings：持仓或资产配置weekly_events：本周事件列表market_view：市场观点/重点事件communication_logs：沟通记录todos：待办事项constraints：限制条件或合规约束输出结构
summary：周度摘要key_points：重点事项risk_notes：风险提示next_actions：后续动作与责任人open_questions：待确认事项markdown_report：完整周报正文| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,717 | 33,397 | +69% | 1 | 1 | 0% | 2,869 | 3,858 | +34% | 0 | 0 | — |
case-02 | pass→pass | 17,083 | 23,557 | +38% | 1 | 1 | 0% | 2,337 | 3,975 | +70% | 0 | 0 | — |
case-03 | fail→fail | 22,375 | 17,918 | -20% | 1 | 1 | 0% | 2,846 | 3,771 | +33% | 0 | 0 | — |
case-04 | pass→pass | 19,563 | 21,141 | +8% | 1 | 1 | 0% | 2,602 | 3,469 | +33% | 0 | 0 | — |
case-05 | fail→pass | 13,689 | 16,981 | +24% | 1 | 1 | 0% | 2,128 | 3,640 | +71% | 0 | 0 | — |
case-06 | pass→pass | 14,102 | 21,684 | +54% | 1 | 1 | 0% | 2,291 | 3,861 | +69% | 0 | 0 | — |
case-07 | pass→pass | 18,858 | 25,264 | +34% | 1 | 1 | 0% | 2,397 | 4,664 | +95% | 0 | 0 | — |
case-08 | pass→pass | 17,388 | 17,901 | +3% | 1 | 1 | 0% | 2,079 | 3,414 | +64% | 0 | 0 | — |
case-09 | pass→pass | 45,618 | 22,403 | -51% | 1 | 1 | 0% | 2,902 | 4,040 | +39% | 0 | 0 | — |
case-10 | fail→pass | 18,521 | 18,563 | +0% | 1 | 1 | 0% | 2,314 | 4,012 | +73% | 0 | 0 | — |
case-11 | fail→pass | 15,934 | 10,795 | -32% | 1 | 1 | 0% | 3,133 | 2,580 | -18% | 0 | 0 | — |
case-12 | fail→pass | 17,124 | 18,885 | +10% | 1 | 1 | 0% | 2,973 | 3,831 | +29% | 0 | 0 | — |
case-13 | pass→pass | 14,149 | 18,134 | +28% | 1 | 1 | 0% | 2,104 | 3,754 | +78% | 0 | 0 | — |
case-14 | pass→pass | 15,613 | 16,725 | +7% | 1 | 1 | 0% | 2,288 | 3,453 | +51% | 0 | 0 | — |
case-15 | pass→pass | 13,172 | 18,841 | +43% | 1 | 1 | 0% | 1,954 | 3,879 | +99% | 0 | 0 | — |
case-16 | pass→pass | 10,503 | 13,548 | +29% | 1 | 1 | 0% | 1,609 | 2,713 | +69% | 0 | 0 | — |
case-17 | pass→pass | 20,312 | 22,322 | +10% | 1 | 1 | 0% | 2,658 | 4,554 | +71% | 0 | 0 | — |
case-18 | pass→pass | 24,990 | 21,623 | -13% | 1 | 1 | 0% | 3,010 | 3,620 | +20% | 0 | 0 | — |
case-19 | pass→pass | 16,260 | 20,228 | +24% | 1 | 1 | 0% | 2,337 | 4,073 | +74% | 0 | 0 | — |
case-20 | pass→pass | 13,695 | 16,581 | +21% | 1 | 1 | 0% | 1,962 | 3,155 | +61% | 0 | 0 | — |
case-21 | pass→pass | 26,054 | 24,650 | -5% | 1 | 1 | 0% | 3,282 | 4,429 | +35% | 0 | 0 | — |
case-22 | fail→pass | 16,550 | 20,381 | +23% | 1 | 1 | 0% | 2,444 | 3,874 | +59% | 0 | 0 | — |
case-23 | pass→pass | 38,667 | 18,137 | -53% | 1 | 1 | 0% | 1,762 | 3,058 | +74% | 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 +26 percentage points is the difference between those two pass rates over the 23 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.