Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股市场深度/订单簿分析。当用户说"市场深度"、"market depth"、"订单簿"、"order book"、"盘口深度"、"挂单量"、"委托深度"时触发。基于 cn-stock-data 获取数据，分析市场深度与订单簿结构。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-market-depth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-06 | ✓→✗ | ▼ Worse | -61% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 43% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 市场深度分析报告
## 一、深度概览
| 档位 | 买方(万股) | 卖方(万股) | 不平衡 |
|------|-----------|-----------|--------|
## 二、订单簿形态
[形态判断与含义]
## 三、深度动态
[恢复速度、异常事件]
## 四、交易建议## [标的] 深度速览
- 买方5档累计 12万股，卖方 8万股
- 深度不平衡 +0.20，买方占优
- 订单簿形态：买厚型，支撑较强
- 建议：25.20有强支撑(大单挂买)参考 references/market-depth-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 | 33,248 | 11,160 | -66% | 1 | 1 | 0% | 5,297 | 1,252 | -76% | 0 | 0 | — |
case-02 | fail→pass | 19,477 | 14,502 | -26% | 1 | 1 | 0% | 3,360 | 2,815 | -16% | 0 | 0 | — |
case-03 | pass→pass | 33,469 | 41,673 | +25% | 1 | 1 | 0% | 5,038 | 7,229 | +43% | 0 | 0 | — |
case-04 | pass→pass | 28,388 | 39,778 | +40% | 1 | 1 | 0% | 5,539 | 8,391 | +51% | 0 | 0 | — |
case-05 | pass→pass | 22,130 | 25,793 | +17% | 1 | 1 | 0% | 3,361 | 4,271 | +27% | 0 | 0 | — |
case-06 | pass→fail | 22,199 | 8,610 | -61% | 1 | 1 | 0% | 3,257 | 1,259 | -61% | 0 | 0 | — |
case-07 | pass→pass | 18,697 | 19,576 | +5% | 1 | 1 | 0% | 3,453 | 3,985 | +15% | 0 | 0 | — |
case-08 | pass→pass | 19,774 | 12,385 | -37% | 1 | 1 | 0% | 2,940 | 2,642 | -10% | 0 | 0 | — |
case-09 | pass→pass | 26,282 | 20,945 | -20% | 1 | 1 | 0% | 3,343 | 3,323 | -1% | 0 | 0 | — |
case-10 | fail→pass | 18,670 | 15,904 | -15% | 1 | 1 | 0% | 2,755 | 2,711 | -2% | 0 | 0 | — |
case-11 | pass→pass | 15,265 | 27,056 | +77% | 1 | 1 | 0% | 2,159 | 1,779 | -18% | 0 | 0 | — |
case-12 | fail→pass | 19,518 | 18,370 | -6% | 1 | 1 | 0% | 2,762 | 3,065 | +11% | 0 | 0 | — |
case-13 | pass→pass | 21,010 | 20,953 | -0% | 1 | 1 | 0% | 3,562 | 3,869 | +9% | 0 | 0 | — |
case-14 | pass→pass | 14,903 | 17,956 | +20% | 1 | 1 | 0% | 2,146 | 3,334 | +55% | 0 | 0 | — |
case-15 | pass→pass | 20,289 | 20,947 | +3% | 1 | 1 | 0% | 3,086 | 3,848 | +25% | 0 | 0 | — |
case-16 | pass→pass | 18,948 | 20,329 | +7% | 1 | 1 | 0% | 2,872 | 3,562 | +24% | 0 | 0 | — |
case-17 | fail→fail | 12,104 | 11,422 | -6% | 1 | 1 | 0% | 2,197 | 1,221 | -44% | 0 | 0 | — |
case-18 | fail→fail | 30,468 | 45,335 | +49% | 1 | 1 | 0% | 4,954 | 7,706 | +56% | 0 | 0 | — |
case-19 | pass→pass | 19,114 | 22,205 | +16% | 1 | 1 | 0% | 2,713 | 3,949 | +46% | 0 | 0 | — |
case-20 | pass→pass | 13,666 | 7,953 | -42% | 1 | 1 | 0% | 2,045 | 2,091 | +2% | 0 | 0 | — |
case-21 | pass→pass | 23,653 | 16,776 | -29% | 1 | 1 | 0% | 3,568 | 3,635 | +2% | 0 | 0 | — |
case-22 | pass→pass | 9,484 | 9,121 | -4% | 1 | 1 | 0% | 1,493 | 2,026 | +36% | 0 | 0 | — |
case-23 | pass→pass | 51,559 | 48,469 | -6% | 1 | 1 | 0% | 8,241 | 8,883 | +8% | 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, and 20 counted toward the lift figure. The other 3 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 +9 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | +21% |
Other measured skills in the registry, with their headline benchmark lift.