Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股知情交易/PIN模型分析。当用户说"知情交易"、"PIN"、"informed trading"、"内幕交易"、"信息交易"、"知情交易概率"时触发。基于 cn-stock-data 获取数据，估算知情交易概率。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-informed-trading/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 61% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 44% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 13% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 6% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 知情交易分析报告
## 一、PIN估计
| 参数 | 数值 |
|------|------|
| PIN | 0.15 |
## 二、VPIN监控
[VPIN时序与预警]
## 三、事件分析
[公告前后PIN变化]
## 四、风险提示## [标的] 知情交易速览
- PIN = 0.15，知情交易概率中等
- VPIN近期稳定，无异常
- 近期无公告前PIN异常上升
- 信息不对称程度：正常参考 references/informed-trading-guide.md 获取详细方法论与 A股实证研究。
python# 调用 skill result = run_skill({ "param1": "value1", "param2": "value2" })
bashpython scripts/run_skill.py --input data.json
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 19,639 | 33,727 | +72% | 1 | 1 | 0% | 3,399 | 5,476 | +61% | 0 | 0 | — |
case-02 | pass→pass | 20,194 | 25,268 | +25% | 1 | 1 | 0% | 4,114 | 5,925 | +44% | 0 | 0 | — |
case-03 | pass→pass | 26,986 | 33,287 | +23% | 1 | 1 | 0% | 5,307 | 5,985 | +13% | 0 | 0 | — |
case-04 | fail→fail | 30,795 | 29,183 | -5% | 1 | 1 | 0% | 4,758 | 4,855 | +2% | 0 | 0 | — |
case-05 | fail→fail | 18,553 | 15,110 | -19% | 1 | 1 | 0% | 2,535 | 2,873 | +13% | 0 | 0 | — |
case-06 | fail→fail | 31,487 | 46,987 | +49% | 1 | 1 | 0% | 5,292 | 7,449 | +41% | 0 | 0 | — |
case-07 | pass→pass | 22,138 | 19,091 | -14% | 1 | 1 | 0% | 4,002 | 4,243 | +6% | 0 | 0 | — |
case-08 | pass→pass | 23,736 | 26,247 | +11% | 1 | 1 | 0% | 3,234 | 4,231 | +31% | 0 | 0 | — |
case-09 | pass→pass | 24,759 | 26,021 | +5% | 1 | 1 | 0% | 3,255 | 4,742 | +46% | 0 | 0 | — |
case-10 | pass→pass | 25,236 | 23,685 | -6% | 1 | 1 | 0% | 3,609 | 4,324 | +20% | 0 | 0 | — |
case-11 | pass→pass | 24,175 | 28,530 | +18% | 1 | 1 | 0% | 3,396 | 4,365 | +29% | 0 | 0 | — |
case-12 | pass→pass | 27,709 | 19,914 | -28% | 1 | 1 | 0% | 2,641 | 3,639 | +38% | 0 | 0 | — |
case-13 | pass→pass | 22,071 | 28,023 | +27% | 1 | 1 | 0% | 3,161 | 4,746 | +50% | 0 | 0 | — |
case-14 | fail→fail | 21,761 | 8,902 | -59% | 1 | 1 | 0% | 2,711 | 1,953 | -28% | 0 | 0 | — |
case-15 | fail→fail | 34,184 | 27,708 | -19% | 1 | 1 | 0% | 5,062 | 5,098 | +1% | 0 | 0 | — |
case-16 | fail→fail | 21,529 | 18,285 | -15% | 1 | 1 | 0% | 3,755 | 3,673 | -2% | 0 | 0 | — |
case-17 | fail→pass | 18,575 | 13,380 | -28% | 1 | 1 | 0% | 2,781 | 2,769 | -0% | 0 | 0 | — |
case-18 | pass→pass | 19,202 | 19,570 | +2% | 1 | 1 | 0% | 2,777 | 3,628 | +31% | 0 | 0 | — |
case-19 | pass→pass | 13,842 | 13,198 | -5% | 1 | 1 | 0% | 2,409 | 3,027 | +26% | 0 | 0 | — |
case-20 | fail→fail | 19,684 | 10,950 | -44% | 1 | 1 | 0% | 2,476 | 2,181 | -12% | 0 | 0 | — |
case-21 | pass→pass | 19,628 | 21,607 | +10% | 1 | 1 | 0% | 3,127 | 4,188 | +34% | 0 | 0 | — |
case-22 | pass→pass | 23,024 | 27,382 | +19% | 1 | 1 | 0% | 3,807 | 4,521 | +19% | 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 +5 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.