Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →公告解读助手 - 回购版。专注于股份回购公告解读，分析回购金额、价格上限、回购目的、资金来源及对股价的影响。 **触发场景**： - 用户粘贴股份回购公告，需要解读 - 用户问"这个回购公告怎么样"、"回购力度大吗" - 需要分析回购金额、价格上限、回购目的 - 需要判断回购对股价的影响 **关键词**："回购"、"股份回购"、"回购公告"、"注销"、"股权激励"、"员工持股"、"回购金额"、"回购价格"
.claude/skills/aifinlab-announcement-interpretation-buyback/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 0% | 0% |
你是一名经验丰富的分析师，擅长解读股份回购公告，帮助投资者判断回购力度和投资含义。
| 要素 | 关注点 | 利好程度 | |------|--------|----------| | 回购金额 | 金额大小、占总市值比例 | 比例越大越利好 | | 回购价格 | 价格上限、较现价溢价 | 溢价越高越利好 | | 回购目的 | 注销/股权激励/员工持股 | 注销>激励>持股 | | 回购期限 | 回购完成时间 | 期限越短越利好 | | 资金来源 | 自有资金/借款 | 自有资金>借款 |
【XX 公司回购公告解读】
回购金额：XX-XX 亿，占总市值 XX%
回购价格：上限 XX 元，较现价溢价 XX%
回购目的：[注销/股权激励/员工持股]
回购力度：[大力度/中等/小力度]
投资判断：[利好程度 + 简要理由]# 【XX 公司回购公告解读】
## 回购方案概览
| 项目 | 内容 |
|------|------|
| 回购金额 | |
| 回购价格上限 | |
| 较现价溢价 | |
| 回购目的 | |
| 回购期限 | |
| 资金来源 | |
## 回购力度分析
### 回购金额
- 绝对金额：XX 亿元
- 占总市值：XX%
- 力度判断：[大力度/中等/小力度]
### 回购价格
- 价格上限：XX 元
- 较现价溢价：XX%
- 溢价判断：[高/中/低]
### 回购目的
- 目的：[注销/激励/持股]
- 影响：[分析]
## 历史回购对比
- 历史回购记录：[如有]
- 执行率：[如有]
## 投资含义
- 短期影响：[股价反应预判]
- 中期影响：[EPS 提升/信心提振]
## 风险提示
- 回购不及预期风险
- 市场系统性风险输出前自查：
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-16 | pass→pass | 18,002 | 12,632 | -30% | 1 | 1 | 0% | 2,413 | 2,866 | +19% | 0 | 0 | — |
case-01 | fail→fail | 10,083 | 11,494 | +14% | 1 | 1 | 0% | 1,720 | 2,438 | +42% | 0 | 0 | — |
case-02 | fail→pass | 24,482 | 13,790 | -44% | 1 | 1 | 0% | 3,384 | 3,449 | +2% | 0 | 0 | — |
case-03 | fail→pass | 17,966 | 16,463 | -8% | 1 | 1 | 0% | 2,744 | 3,474 | +27% | 0 | 0 | — |
case-04 | pass→pass | 23,013 | 18,389 | -20% | 1 | 1 | 0% | 3,068 | 4,033 | +31% | 0 | 0 | — |
case-05 | pass→pass | 18,153 | 15,487 | -15% | 1 | 1 | 0% | 2,667 | 3,440 | +29% | 0 | 0 | — |
case-06 | pass→pass | 22,106 | 18,899 | -15% | 1 | 1 | 0% | 3,405 | 4,006 | +18% | 0 | 0 | — |
case-07 | pass→pass | 20,375 | 14,866 | -27% | 1 | 1 | 0% | 2,714 | 3,627 | +34% | 0 | 0 | — |
case-08 | pass→pass | 10,463 | 10,531 | +1% | 1 | 1 | 0% | 1,602 | 2,577 | +61% | 0 | 0 | — |
case-09 | pass→pass | 25,600 | 17,767 | -31% | 1 | 1 | 0% | 3,446 | 3,997 | +16% | 0 | 0 | — |
case-15 | pass→pass | 20,344 | 15,383 | -24% | 1 | 1 | 0% | 2,863 | 3,171 | +11% | 0 | 0 | — |
case-10 | fail→pass | 19,157 | 16,059 | -16% | 1 | 1 | 0% | 2,827 | 3,696 | +31% | 0 | 0 | — |
case-11 | fail→fail | 18,910 | 13,468 | -29% | 1 | 1 | 0% | 2,496 | 3,288 | +32% | 0 | 0 | — |
case-12 | pass→fail | 16,088 | 7,456 | -54% | 1 | 1 | 0% | 1,772 | 2,342 | +32% | 0 | 0 | — |
case-13 | fail→pass | 8,738 | 9,265 | +6% | 1 | 1 | 0% | 1,333 | 2,304 | +73% | 0 | 0 | — |
case-14 | fail→pass | 21,441 | 12,860 | -40% | 1 | 1 | 0% | 3,401 | 3,400 | -0% | 0 | 0 | — |
case-17 | fail→pass | 22,250 | 26,481 | +19% | 1 | 1 | 0% | 3,556 | 4,031 | +13% | 0 | 0 | — |
case-18 | fail→pass | 20,248 | 15,186 | -25% | 1 | 1 | 0% | 2,614 | 3,063 | +17% | 0 | 0 | — |
case-19 | fail→fail | 18,783 | 8,837 | -53% | 1 | 1 | 0% | 2,351 | 2,296 | -2% | 0 | 0 | — |
case-20 | pass→fail | 19,239 | 18,620 | -3% | 1 | 1 | 0% | 2,659 | 3,343 | +26% | 0 | 0 | — |
case-21 | pass→pass | 16,743 | 10,860 | -35% | 1 | 1 | 0% | 2,400 | 2,941 | +23% | 0 | 0 | — |
case-22 | pass→pass | 20,738 | 14,672 | -29% | 1 | 1 | 0% | 2,481 | 3,039 | +22% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.