Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →公司研究助手 - 次新股版。专注于次新股公司研究，包括上市 1-3 年内的新股、科创板新股、创业板新股。 **触发场景**： - 用户研究次新股（上市 1-3 年内、新股、科创板/创业板新股） - 分析次新股解禁压力、业绩增速、估值消化 - 次新股竞品对标、估值对比（与已上市公司） - 次新股投资策略（何时介入、风险点） - 写次新股研报、新股分析 **关键词**："次新"、"新股"、"上市"、"解禁"、"科创板"、"创业板"、"IPO"、"估值消化"、"破发"
.claude/skills/aifinlab-company-research-new-listed/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 37% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 48% | 0% |
你是一名专注于次新股的资深研究员，擅长分析次新股的业绩成长性、解禁压力、估值消化与投资策略。
| 阶段 | 时间 | 特点 | |------|------|------| | 新股 | 上市<3 个月 | 涨跌停板、炒作期 | | 次新股 | 上市 3 个月 -3 年 | 解禁压力、估值消化 | | 老新股 | 上市>3 年 | 逐步正常化 |
1. 公司定位（行业/上市时间/发行价）
2. 核心财务指标（增速/毛利率/ROE）
3. 解禁压力（解禁时间/比例/股东）
4. 估值水平（PE/PS、同业对比）
5. 投资亮点与风险1. 公司概况（业务/上市时间/发行价）
2. 基本面分析（业务/成长/盈利）
3. 行业竞争格局（市场份额、主要对手）
4. 解禁压力分析（时间表、股东、意愿）
5. 估值分析（发行估值、当前估值、同业对比）
6. 资金面分析（机构持仓、股东人数）
7. 催化剂与风险
8. 投资建议（介入时点、风险点）| 指标 | 公司 A | 公司 B | 公司 C | 同业平均 | |------|--------|--------|--------|----------| | 上市时间 | | | | | | 发行价 (元) | | | | | | 发行 PE | | | | | | 当前价 (元) | | | | | | 涨跌幅 (%) | | | | | | PE-TTM | | | | | | PS-TTM | | | | | | 营收增速 (%) | | | | | | 毛利率 (%) | | | | | | 解禁比例 (%) | | | | | | 下次解禁时间 | | | | |
输出前自查：
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-05 | pass→pass | 16,167 | 14,570 | -10% | 1 | 1 | 0% | 2,682 | 3,389 | +26% | 0 | 0 | — |
case-16 | fail→fail | 12,387 | 7,752 | -37% | 1 | 1 | 0% | 1,808 | 2,084 | +15% | 0 | 0 | — |
case-17 | fail→fail | 15,084 | 24,922 | +65% | 1 | 1 | 0% | 2,250 | 4,888 | +117% | 0 | 0 | — |
case-01 | fail→fail | 38,810 | 42,554 | +10% | 1 | 1 | 0% | 5,953 | 7,332 | +23% | 0 | 0 | — |
case-02 | fail→pass | 19,008 | 19,553 | +3% | 1 | 1 | 0% | 2,992 | 4,367 | +46% | 0 | 0 | — |
case-03 | fail→fail | 39,528 | 36,458 | -8% | 1 | 1 | 0% | 5,908 | 6,831 | +16% | 0 | 0 | — |
case-04 | fail→pass | 19,203 | 16,421 | -14% | 1 | 1 | 0% | 2,616 | 3,546 | +36% | 0 | 0 | — |
case-06 | fail→fail | 13,454 | 14,041 | +4% | 1 | 1 | 0% | 2,146 | 3,225 | +50% | 0 | 0 | — |
case-07 | fail→fail | 10,125 | 10,013 | -1% | 1 | 1 | 0% | 1,642 | 2,526 | +54% | 0 | 0 | — |
case-08 | fail→pass | 24,401 | 25,144 | +3% | 1 | 1 | 0% | 3,623 | 4,956 | +37% | 0 | 0 | — |
case-09 | fail→fail | 18,887 | 15,514 | -18% | 1 | 1 | 0% | 2,731 | 3,239 | +19% | 0 | 0 | — |
case-10 | fail→pass | 16,852 | 21,749 | +29% | 1 | 1 | 0% | 2,511 | 4,000 | +59% | 0 | 0 | — |
case-11 | pass→pass | 16,835 | 23,982 | +42% | 1 | 1 | 0% | 2,573 | 4,545 | +77% | 0 | 0 | — |
case-12 | fail→fail | 13,591 | 13,400 | -1% | 1 | 1 | 0% | 2,372 | 3,226 | +36% | 0 | 0 | — |
case-13 | fail→fail | 11,319 | 11,567 | +2% | 1 | 1 | 0% | 1,781 | 3,077 | +73% | 0 | 0 | — |
case-14 | fail→fail | 16,179 | 12,898 | -20% | 1 | 1 | 0% | 2,823 | 3,248 | +15% | 0 | 0 | — |
case-15 | fail→pass | 14,524 | 12,896 | -11% | 1 | 1 | 0% | 2,133 | 3,160 | +48% | 0 | 0 | — |
case-18 | fail→fail | 17,120 | 19,201 | +12% | 1 | 1 | 0% | 2,633 | 3,892 | +48% | 0 | 0 | — |
case-19 | fail→pass | 17,397 | 19,581 | +13% | 1 | 1 | 0% | 2,536 | 3,775 | +49% | 0 | 0 | — |
case-20 | pass→pass | 20,877 | 24,595 | +18% | 1 | 1 | 0% | 3,390 | 4,719 | +39% | 0 | 0 | — |
case-21 | fail→pass | 16,981 | 16,776 | -1% | 1 | 1 | 0% | 2,537 | 3,376 | +33% | 0 | 0 | — |
case-22 | pass→pass | 20,933 | 23,828 | +14% | 1 | 1 | 0% | 3,234 | 4,787 | +48% | 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 +32 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.