Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →公司研究助手 - 成长股版。专注于成长型公司研究，包括高增速、高研发投入、新兴赛道的公司。 **触发场景**： - 用户研究成长型公司（营收增速>30%、新兴赛道、高研发投入） - 分析成长公司增长驱动、市场空间、渗透率 - 成长公司估值分析（PS、PEG、市值空间） - 成长公司竞品对标、成长持续性分析 - 写成长公司研报、投资建议 **关键词**："成长"、"高增长"、"增速"、"渗透率"、"市场空间"、"研发投入"、"PS"、"PEG"、"新兴赛道"、"科技创新"
.claude/skills/aifinlab-company-research-growth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 57% | 0% |
你是一名专注于成长型公司的资深研究员，擅长分析增长驱动、市场空间、竞争格局与成长持续性。
| 标准 | 指标 | |------|------| | 营收增速 | >30%（近 3 年 CAGR） | | 行业赛道 | 新兴赛道、高增长行业 | | 研发投入 | 研发费用率>10% | | 市场空间 | 渗透率<50%、天花板高 |
1. 公司定位（赛道/增速/渗透率）
2. 增长驱动（市场空间/渗透率/新品）
3. 核心财务指标（增速/毛利率/研发）
4. 估值水平（PS/PEG、市值空间）
5. 投资亮点与风险1. 公司概况（业务/赛道/定位）
2. 市场空间测算（TAM/SAM/SOM）
3. 增长驱动分析（量/价/新品/新市场）
4. 竞争力分析（技术/产品/客户/渠道）
5. 财务分析（成长/盈利/现金流）
6. 竞争格局（主要对手、市占率）
7. 估值分析（PS/PEG、市值空间）
8. 催化剂与风险
9. 投资建议| 指标 | 公司 A | 公司 B | 公司 C | 行业平均 | |------|--------|--------|--------|----------| | 市值 (亿) | | | | | | PS-TTM | | | | | | PEG | | | | | | 营收增速 (%) | | | | | | 研发费用率 (%) | | | | | | 毛利率 (%) | | | | | | 渗透率 (%) | | | | | | 市场空间 (亿) | | | | |
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 | 18,508 | 18,573 | +0% | 1 | 1 | 0% | 2,694 | 3,742 | +39% | 0 | 0 | — |
case-13 | fail→fail | 16,384 | 18,020 | +10% | 1 | 1 | 0% | 2,462 | 3,687 | +50% | 0 | 0 | — |
case-12 | fail→fail | 18,373 | 18,853 | +3% | 1 | 1 | 0% | 2,740 | 3,629 | +32% | 0 | 0 | — |
case-14 | fail→fail | 18,346 | 21,564 | +18% | 1 | 1 | 0% | 2,554 | 4,156 | +63% | 0 | 0 | — |
case-04 | pass→pass | 15,145 | 13,375 | -12% | 1 | 1 | 0% | 2,431 | 3,051 | +26% | 0 | 0 | — |
case-01 | fail→pass | 17,468 | 25,945 | +49% | 1 | 1 | 0% | 2,770 | 4,335 | +56% | 0 | 0 | — |
case-02 | fail→pass | 37,701 | 32,531 | -14% | 1 | 1 | 0% | 5,839 | 6,252 | +7% | 0 | 0 | — |
case-03 | fail→pass | 25,962 | 37,867 | +46% | 1 | 1 | 0% | 4,133 | 7,128 | +72% | 0 | 0 | — |
case-06 | pass→pass | 12,649 | 11,823 | -7% | 1 | 1 | 0% | 2,114 | 2,930 | +39% | 0 | 0 | — |
case-07 | fail→pass | 9,268 | 4,184 | -55% | 1 | 1 | 0% | 1,548 | 1,667 | +8% | 0 | 0 | — |
case-08 | fail→pass | 12,015 | 14,027 | +17% | 1 | 1 | 0% | 1,908 | 2,987 | +57% | 0 | 0 | — |
case-09 | pass→fail | 21,021 | 25,057 | +19% | 1 | 1 | 0% | 3,311 | 4,739 | +43% | 0 | 0 | — |
case-10 | fail→pass | 12,377 | 13,330 | +8% | 1 | 1 | 0% | 1,980 | 3,030 | +53% | 0 | 0 | — |
case-11 | fail→fail | 10,982 | 7,631 | -31% | 1 | 1 | 0% | 1,764 | 2,206 | +25% | 0 | 0 | — |
case-15 | fail→pass | 17,309 | 16,835 | -3% | 1 | 1 | 0% | 2,906 | 3,778 | +30% | 0 | 0 | — |
case-16 | pass→pass | 13,767 | 24,807 | +80% | 1 | 1 | 0% | 2,018 | 2,415 | +20% | 0 | 0 | — |
case-17 | fail→pass | 10,511 | 2,994 | -72% | 1 | 1 | 0% | 1,618 | 1,406 | -13% | 0 | 0 | — |
case-18 | fail→pass | 6,288 | 2,795 | -56% | 1 | 1 | 0% | 1,020 | 1,387 | +36% | 0 | 0 | — |
case-19 | fail→pass | 10,879 | 3,980 | -63% | 1 | 1 | 0% | 1,736 | 1,613 | -7% | 0 | 0 | — |
case-20 | fail→pass | 13,099 | 13,212 | +1% | 1 | 1 | 0% | 1,972 | 2,989 | +52% | 0 | 0 | — |
case-21 | pass→pass | 16,999 | 17,998 | +6% | 1 | 1 | 0% | 2,515 | 3,627 | +44% | 0 | 0 | — |
case-22 | fail→pass | 21,673 | 16,559 | -24% | 1 | 1 | 0% | 3,281 | 3,566 | +9% | 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 +50 percentage points is the difference between those two pass rates over the 22 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.
Other measured skills in the registry, with their headline benchmark lift.