Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股集合竞价/开盘博弈分析。当用户说"集合竞价"、"开盘竞价"、"9:15"、"9:25"、"开盘价"、"竞价分析"、"早盘竞价"、"竞价抢筹"时触发。基于 cn-stock-data 获取数据，分析集合竞价量价特征、资金博弈、开盘信号。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-opening-auction/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-04 | ✓→✗ | ▼ Worse | -62% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 54% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 76% | 0% |
| case-08 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 集合竞价分析报告
## 一、竞价概览
| 指标 | 数值 | 评估 |
|------|------|------|
| 开盘价 | 25.50 | 高开1.2% |
| 竞价量 | 85万股 | 放量 |
## 二、资金博弈
[大单方向、委托变化]
## 三、开盘信号
[信号类型、历史统计]
## 四、操作建议## [标的] 竞价速览
- 高开 +1.2%，竞价量85万股(日均3.5%)
- 9:20后大单净买入 +320万
- 信号：高开放量，偏强
- 建议：开盘观察5分钟确认方向参考 references/opening-auction-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 | fail→fail | 77,727 | 29,247 | -62% | 1 | 1 | 0% | 6,009 | 4,666 | -22% | 0 | 0 | — |
case-02 | fail→fail | 18,092 | 13,458 | -26% | 1 | 1 | 0% | 2,421 | 1,297 | -46% | 0 | 0 | — |
case-03 | fail→fail | 26,120 | 38,676 | +48% | 1 | 1 | 0% | 3,819 | 5,967 | +56% | 0 | 0 | — |
case-04 | pass→fail | 24,027 | 11,596 | -52% | 1 | 1 | 0% | 3,443 | 1,320 | -62% | 0 | 0 | — |
case-05 | pass→pass | 23,081 | 29,706 | +29% | 1 | 1 | 0% | 3,271 | 5,037 | +54% | 0 | 0 | — |
case-06 | pass→pass | 19,878 | 37,632 | +89% | 1 | 1 | 0% | 2,530 | 4,441 | +76% | 0 | 0 | — |
case-07 | fail→pass | 11,805 | 10,588 | -10% | 1 | 1 | 0% | 1,640 | 2,175 | +33% | 0 | 0 | — |
case-08 | pass→pass | 27,535 | 22,378 | -19% | 1 | 1 | 0% | 3,470 | 4,054 | +17% | 0 | 0 | — |
case-09 | pass→pass | 33,207 | 17,996 | -46% | 1 | 1 | 0% | 2,506 | 3,063 | +22% | 0 | 0 | — |
case-10 | pass→pass | 20,460 | 22,308 | +9% | 1 | 1 | 0% | 2,605 | 3,695 | +42% | 0 | 0 | — |
case-11 | pass→pass | 19,791 | 21,569 | +9% | 1 | 1 | 0% | 2,795 | 3,413 | +22% | 0 | 0 | — |
case-12 | fail→fail | 19,543 | 20,576 | +5% | 1 | 1 | 0% | 2,553 | 3,447 | +35% | 0 | 0 | — |
case-13 | fail→fail | 11,281 | 20,097 | +78% | 1 | 1 | 0% | 1,612 | 2,811 | +74% | 0 | 0 | — |
case-14 | fail→fail | 17,125 | 15,895 | -7% | 1 | 1 | 0% | 2,499 | 3,358 | +34% | 0 | 0 | — |
case-15 | pass→pass | 25,579 | 27,890 | +9% | 1 | 1 | 0% | 3,205 | 4,275 | +33% | 0 | 0 | — |
case-16 | pass→pass | 21,783 | 22,552 | +4% | 1 | 1 | 0% | 3,199 | 3,437 | +7% | 0 | 0 | — |
case-17 | pass→pass | 43,497 | 26,260 | -40% | 1 | 1 | 0% | 3,255 | 4,037 | +24% | 0 | 0 | — |
case-18 | fail→fail | 15,017 | 9,380 | -38% | 1 | 1 | 0% | 2,223 | 2,231 | +0% | 0 | 0 | — |
case-19 | pass→pass | 20,988 | 22,022 | +5% | 1 | 1 | 0% | 2,658 | 3,633 | +37% | 0 | 0 | — |
case-20 | pass→pass | 21,012 | 13,895 | -34% | 1 | 1 | 0% | 2,515 | 2,774 | +10% | 0 | 0 | — |
case-21 | fail→fail | 65,274 | 25,201 | -61% | 1 | 1 | 0% | 3,336 | 3,919 | +17% | 0 | 0 | — |
case-22 | fail→fail | 7,644 | 6,968 | -9% | 1 | 1 | 0% | 1,093 | 1,706 | +56% | 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, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of 0 percentage points is the difference between those two pass rates over the 20 comparable cases. 3 cases got worse with the skill loaded, and they are 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.