Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股龙虎榜分析/游资追踪。当用户说"龙虎榜"、"游资"、"营业部"、"席位"、"涨停分析"、"dragon tiger"、"谁在买"、"机构席位"、"游资大佬"、"XX上龙虎榜了吗"时触发。基于 efinance 龙虎榜数据，分析个股上榜原因、买卖席位、机构vs游资博弈、近期龙虎榜活跃股。支持券商研报风格（formal）和快速解读风格（brief）。不适用于北向资金分析（用 a-share-northbound）或基金持仓（用 a-share-fund-holding）。
.claude/skills/aifinlab-a-share-dragon-tiger/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -45% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -1% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" # 龙虎榜数据（通过 akshare，cn-stock-data 不支持此命令） # 方式1: akshare 龙虎榜明细（推荐，数据最全） python -c "import akshare as ak; df=ak.stock_lhb_detail_em(start_date='YYYYMMDD', end_date='YYYYMMDD'); print(df.to_json(orient='records', force_ascii=False))" # 方式2: efinance 龙虎榜（备选） python -c "import efinance as ef; df=ef.stock.get_daily_billboard(); print(df.to_json(orient='records', force_ascii=False))" # 个股行情（通过 cn-stock-data） python "$SCRIPTS/cn_stock_data.py" quote --code [CODE] # K线（看上榜前后走势） python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [日期] # 资金流向 python "$SCRIPTS/cn_stock_data.py" fund_flow --code [CODE]
补充：通过 web 搜索确认席位背后的游资/机构身份。
根据用户意图分类：
获取龙虎榜明细（买卖前5席位、金额、占比）+ 个股行情 + 近期K线。多只股票时并行获取。
> 详见 references/dragon-tiger-guide.md 中的席位类型识别和常见模式。
根据风格生成报告：
| 维度 | formal | brief | |------|--------|-------| | 输出格式 | 完整龙虎榜报告 | 快速要点 | | 席位分析 | 逐席详细分析 | 仅标注机构/知名游资 | | 统计 | 净买卖汇总+历史上榜统计 | 仅净买卖方向 | | 结论 | 客观陈述（"机构净买入X亿"） | 可加判断 |
默认风格：brief。用户要求"详细分析"/"出报告"时用 formal。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 30,121 | 15,060 | -50% | 1 | 1 | 0% | 3,959 | 1,671 | -58% | 0 | 0 | — |
case-02 | fail→fail | 23,232 | 13,597 | -41% | 1 | 1 | 0% | 2,857 | 1,349 | -53% | 0 | 0 | — |
case-03 | fail→fail | 18,503 | 18,064 | -2% | 1 | 1 | 0% | 2,306 | 1,744 | -24% | 0 | 0 | — |
case-04 | pass→fail | 20,432 | 13,674 | -33% | 1 | 1 | 0% | 3,143 | 1,636 | -48% | 0 | 0 | — |
case-05 | pass→fail | 28,607 | 11,418 | -60% | 1 | 1 | 0% | 4,585 | 1,386 | -70% | 0 | 0 | — |
case-06 | pass→fail | 26,175 | 15,092 | -42% | 1 | 1 | 0% | 3,543 | 1,610 | -55% | 0 | 0 | — |
case-07 | fail→fail | 7,845 | 12,608 | +61% | 1 | 1 | 0% | 981 | 1,578 | +61% | 0 | 0 | — |
case-08 | fail→fail | 29,925 | 11,894 | -60% | 1 | 1 | 0% | 4,204 | 1,503 | -64% | 0 | 0 | — |
case-09 | pass→pass | 13,681 | 10,259 | -25% | 1 | 1 | 0% | 2,498 | 2,662 | +7% | 0 | 0 | — |
case-10 | fail→pass | 18,405 | 4,979 | -73% | 1 | 1 | 0% | 2,863 | 1,572 | -45% | 0 | 0 | — |
case-11 | pass→pass | 14,338 | 15,045 | +5% | 1 | 1 | 0% | 2,201 | 2,885 | +31% | 0 | 0 | — |
case-12 | fail→pass | 22,245 | 17,037 | -23% | 1 | 1 | 0% | 2,912 | 3,335 | +15% | 0 | 0 | — |
case-13 | fail→pass | 28,659 | 17,779 | -38% | 1 | 1 | 0% | 2,136 | 3,176 | +49% | 0 | 0 | — |
case-14 | pass→pass | 19,752 | 15,283 | -23% | 1 | 1 | 0% | 2,662 | 3,184 | +20% | 0 | 0 | — |
case-15 | pass→pass | 19,664 | 18,430 | -6% | 1 | 1 | 0% | 2,813 | 3,208 | +14% | 0 | 0 | — |
case-16 | pass→pass | 14,786 | 16,292 | +10% | 1 | 1 | 0% | 1,961 | 2,986 | +52% | 0 | 0 | — |
case-17 | pass→pass | 22,900 | 18,899 | -17% | 1 | 1 | 0% | 2,768 | 3,432 | +24% | 0 | 0 | — |
case-18 | fail→pass | 22,768 | 15,609 | -31% | 1 | 1 | 0% | 3,581 | 3,625 | +1% | 0 | 0 | — |
case-19 | fail→pass | 20,434 | 13,979 | -32% | 1 | 1 | 0% | 2,697 | 2,670 | -1% | 0 | 0 | — |
case-20 | pass→pass | 21,765 | 12,941 | -41% | 1 | 1 | 0% | 3,052 | 2,430 | -20% | 0 | 0 | — |
case-21 | fail→pass | 21,952 | 8,036 | -63% | 1 | 1 | 0% | 2,370 | 1,863 | -21% | 0 | 0 | — |
case-22 | fail→pass | 13,353 | 2,218 | -83% | 1 | 1 | 0% | 1,801 | 1,181 | -34% | 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 14 counted toward the lift figure. The other 8 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 +18 percentage points is the difference between those two pass rates over the 14 comparable cases. 5 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.