Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股NLP舆情/文本情感量化分析。当用户说"NLP"、"舆情分析"、"文本情感"、"新闻情感"、"sentiment"、"舆情量化"、"研报情感"时触发。基于 cn-stock-data 获取数据，结合NLP分析文本情感与市场关系。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-nlp-sentiment/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 19% | 0% |
通过 cn-stock-data skill 获取数据：
# NLP舆情量化分析报告
## 一、情感概览
| 来源 | 情感分数 | 变化 |
|------|---------|------|
## 二、情感因子
[各情感因子值与分位数]
## 三、事件分析
[近期重大新闻情感影响]
## 四、交易信号
[基于情感的买卖建议]## [标的] 舆情速览
- 新闻情感 +0.35 (偏正面)
- 研报情感：近期2篇上调评级
- 股吧情绪偏乐观(P75)
- 注意：散户过度乐观可能是反向信号参考 references/nlp-sentiment-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 | 29,985 | 29,455 | -2% | 1 | 1 | 0% | 4,518 | 4,333 | -4% | 0 | 0 | — |
case-02 | fail→fail | 14,343 | 9,712 | -32% | 1 | 1 | 0% | 2,081 | 1,915 | -8% | 0 | 0 | — |
case-03 | fail→fail | 25,146 | 37,858 | +51% | 1 | 1 | 0% | 3,972 | 6,366 | +60% | 0 | 0 | — |
case-04 | fail→pass | 19,992 | 24,168 | +21% | 1 | 1 | 0% | 2,862 | 3,783 | +32% | 0 | 0 | — |
case-05 | fail→pass | 20,991 | 8,879 | -58% | 1 | 1 | 0% | 3,487 | 2,210 | -37% | 0 | 0 | — |
case-06 | fail→pass | 20,704 | 26,564 | +28% | 1 | 1 | 0% | 3,503 | 4,418 | +26% | 0 | 0 | — |
case-07 | pass→pass | 20,754 | 19,175 | -8% | 1 | 1 | 0% | 2,684 | 3,325 | +24% | 0 | 0 | — |
case-08 | pass→pass | 22,480 | 19,077 | -15% | 1 | 1 | 0% | 3,273 | 3,368 | +3% | 0 | 0 | — |
case-09 | pass→pass | 19,987 | 24,274 | +21% | 1 | 1 | 0% | 3,220 | 4,080 | +27% | 0 | 0 | — |
case-10 | pass→pass | 19,684 | 19,902 | +1% | 1 | 1 | 0% | 2,975 | 3,682 | +24% | 0 | 0 | — |
case-11 | pass→pass | 24,554 | 17,442 | -29% | 1 | 1 | 0% | 3,299 | 3,327 | +1% | 0 | 0 | — |
case-12 | pass→pass | 12,547 | 14,229 | +13% | 1 | 1 | 0% | 1,883 | 2,834 | +51% | 0 | 0 | — |
case-13 | pass→pass | 23,338 | 26,262 | +13% | 1 | 1 | 0% | 3,339 | 4,702 | +41% | 0 | 0 | — |
case-14 | pass→pass | 12,339 | 5,707 | -54% | 1 | 1 | 0% | 1,936 | 1,587 | -18% | 0 | 0 | — |
case-15 | fail→pass | 14,727 | 10,259 | -30% | 1 | 1 | 0% | 2,310 | 2,177 | -6% | 0 | 0 | — |
case-16 | fail→fail | 21,187 | 16,337 | -23% | 1 | 1 | 0% | 3,203 | 3,099 | -3% | 0 | 0 | — |
case-17 | pass→pass | 18,235 | 14,326 | -21% | 1 | 1 | 0% | 2,526 | 3,004 | +19% | 0 | 0 | — |
case-18 | pass→pass | 21,654 | 20,140 | -7% | 1 | 1 | 0% | 3,355 | 3,831 | +14% | 0 | 0 | — |
case-19 | fail→pass | 21,658 | 23,074 | +7% | 1 | 1 | 0% | 3,297 | 3,935 | +19% | 0 | 0 | — |
case-20 | pass→pass | 12,845 | 18,163 | +41% | 1 | 1 | 0% | 1,879 | 3,013 | +60% | 0 | 0 | — |
case-21 | fail→fail | 21,186 | 15,073 | -29% | 1 | 1 | 0% | 3,563 | 1,729 | -51% | 0 | 0 | — |
case-22 | pass→pass | 20,437 | 36,862 | +80% | 1 | 1 | 0% | 3,541 | 6,509 | +84% | 0 | 0 | — |
case-23 | pass→pass | 24,900 | 30,284 | +22% | 1 | 1 | 0% | 4,795 | 6,289 | +31% | 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 22 counted toward the lift figure. The other 1 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 +22 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | 0% |
Other measured skills in the registry, with their headline benchmark lift.