Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在基金场景中完成投顾组合诊断助手相关任务时使用本技能，适用于组合诊断及相关分析、生成、审阅请求。 用户提到「组合诊断」「投顾组合」「组合体检」或需要输出结构化中文结论、模板和配套脚本时，应优先触发。
.claude/skills/aifinlab-advisory-portfolio-diagnostic-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 97% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 81% | 0% |
本技能用于支持基金业务中的组合诊断工作，重点解决「对投顾组合的收益来源、风险暴露、适配性和持仓结构进行诊断并输出中文结论。」这类任务。输出默认使用中文，要求结构清楚、证据边界明确，并能被投研、产品、销售、投顾或合规团队直接复用。
优先输入包括：
可选增强输入包括：
当输入不完整时，必须先列出缺失材料，再基于现有信息输出阶段性判断，不得编造未确认事实。
建议输出至少覆盖以下模块：
若任务涉及对客材料、话术或解释文案，必须同时输出「不能直接对外使用的前提说明」或「需正式复核」的提醒。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 24,981 | 27,886 | +12% | 1 | 1 | 0% | 3,892 | 4,715 | +21% | 0 | 0 | — |
case-02 | fail→fail | 25,811 | 27,022 | +5% | 1 | 1 | 0% | 3,931 | 5,033 | +28% | 0 | 0 | — |
case-03 | fail→pass | 37,729 | 20,662 | -45% | 1 | 1 | 0% | 3,052 | 3,904 | +28% | 0 | 0 | — |
case-04 | fail→fail | 24,430 | 22,255 | -9% | 1 | 1 | 0% | 3,609 | 4,575 | +27% | 0 | 0 | — |
case-05 | fail→fail | 23,222 | 38,603 | +66% | 1 | 1 | 0% | 4,494 | 7,004 | +56% | 0 | 0 | — |
case-06 | fail→fail | 19,509 | 29,791 | +53% | 1 | 1 | 0% | 2,848 | 5,055 | +77% | 0 | 0 | — |
case-07 | fail→pass | 17,666 | 22,853 | +29% | 1 | 1 | 0% | 2,200 | 4,344 | +97% | 0 | 0 | — |
case-08 | fail→pass | 19,652 | 22,351 | +14% | 1 | 1 | 0% | 2,876 | 4,066 | +41% | 0 | 0 | — |
case-09 | fail→fail | 28,175 | 21,970 | -22% | 1 | 1 | 0% | 3,644 | 3,786 | +4% | 0 | 0 | — |
case-10 | pass→pass | 16,102 | 30,859 | +92% | 1 | 1 | 0% | 2,223 | 5,072 | +128% | 0 | 0 | — |
case-11 | pass→pass | 24,255 | 24,051 | -1% | 1 | 1 | 0% | 3,082 | 4,595 | +49% | 0 | 0 | — |
case-12 | pass→pass | 28,187 | 22,743 | -19% | 1 | 1 | 0% | 3,849 | 4,507 | +17% | 0 | 0 | — |
case-13 | pass→pass | 18,245 | 24,571 | +35% | 1 | 1 | 0% | 2,946 | 4,834 | +64% | 0 | 0 | — |
case-14 | pass→pass | 23,039 | 24,392 | +6% | 1 | 1 | 0% | 3,783 | 4,974 | +31% | 0 | 0 | — |
case-15 | pass→pass | 20,140 | 22,167 | +10% | 1 | 1 | 0% | 3,135 | 4,411 | +41% | 0 | 0 | — |
case-16 | fail→fail | 24,121 | 28,023 | +16% | 1 | 1 | 0% | 3,379 | 4,503 | +33% | 0 | 0 | — |
case-17 | fail→pass | 25,089 | 29,388 | +17% | 1 | 1 | 0% | 3,223 | 5,824 | +81% | 0 | 0 | — |
case-18 | pass→pass | 27,261 | 30,716 | +13% | 1 | 1 | 0% | 3,194 | 5,025 | +57% | 0 | 0 | — |
case-19 | pass→pass | 18,904 | 42,868 | +127% | 1 | 1 | 0% | 2,789 | 4,280 | +53% | 0 | 0 | — |
case-20 | fail→fail | 20,314 | 21,723 | +7% | 1 | 1 | 0% | 3,123 | 4,062 | +30% | 0 | 0 | — |
case-21 | pass→pass | 23,521 | 28,308 | +20% | 1 | 1 | 0% | 3,191 | 5,007 | +57% | 0 | 0 | — |
case-22 | pass→pass | 22,743 | 28,907 | +27% | 1 | 1 | 0% | 3,598 | 4,981 | +38% | 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.