Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股黑天鹅预警/极端事件分析。当用户说"黑天鹅"、"black swan"、"极端事件"、"尾部事件"、"灰犀牛"、"突发风险"、"系统崩溃"时触发。基于 cn-stock-data 获取数据，监控与预警潜在的黑天鹅事件。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-black-swan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 20% | 0% |
通过 cn-stock-data skill 获取数据：
# 黑天鹅预警报告
## 一、预警指标
| 指标 | 数值 | 分位数 | 状态 |
|------|------|--------|------|
## 二、灰犀牛监控
[潜在风险事件评估]
## 三、历史对标
[与历史黑天鹅事件的相似度]
## 四、应对预案
[当前建议的防御措施]## 黑天鹅预警速览
- 波动率P45，正常
- 流动性充足，无异常
- 灰犀牛：房地产风险中等关注
- 状态：绿色，无需特别防御参考 references/black-swan-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-05 | fail→fail | 7,739 | 13,546 | +75% | 1 | 1 | 0% | 1,181 | 2,775 | +135% | 0 | 0 | — |
case-06 | pass→pass | 11,891 | 8,179 | -31% | 1 | 1 | 0% | 1,751 | 2,087 | +19% | 0 | 0 | — |
case-01 | fail→fail | 36,624 | 49,528 | +35% | 1 | 1 | 0% | 5,228 | 7,732 | +48% | 0 | 0 | — |
case-02 | fail→pass | 13,506 | 9,801 | -27% | 1 | 1 | 0% | 2,028 | 2,180 | +7% | 0 | 0 | — |
case-03 | fail→fail | 42,346 | 50,381 | +19% | 1 | 1 | 0% | 5,992 | 7,378 | +23% | 0 | 0 | — |
case-04 | pass→pass | 32,390 | 45,608 | +41% | 1 | 1 | 0% | 5,441 | 7,327 | +35% | 0 | 0 | — |
case-07 | fail→pass | 42,869 | 16,718 | -61% | 1 | 1 | 0% | 3,799 | 3,187 | -16% | 0 | 0 | — |
case-08 | fail→pass | 25,093 | 15,868 | -37% | 1 | 1 | 0% | 2,998 | 3,097 | +3% | 0 | 0 | — |
case-09 | fail→fail | 21,658 | 10,670 | -51% | 1 | 1 | 0% | 2,742 | 2,021 | -26% | 0 | 0 | — |
case-10 | fail→fail | 27,735 | 27,148 | -2% | 1 | 1 | 0% | 3,519 | 3,840 | +9% | 0 | 0 | — |
case-11 | fail→fail | 45,018 | 29,372 | -35% | 1 | 1 | 0% | 5,855 | 4,218 | -28% | 0 | 0 | — |
case-12 | fail→pass | 29,163 | 18,797 | -36% | 1 | 1 | 0% | 4,116 | 3,073 | -25% | 0 | 0 | — |
case-13 | fail→pass | 35,213 | 24,186 | -31% | 1 | 1 | 0% | 3,792 | 4,562 | +20% | 0 | 0 | — |
case-14 | fail→fail | 27,898 | 24,778 | -11% | 1 | 1 | 0% | 3,860 | 4,004 | +4% | 0 | 0 | — |
case-15 | fail→pass | 19,666 | 8,602 | -56% | 1 | 1 | 0% | 2,569 | 1,911 | -26% | 0 | 0 | — |
case-16 | fail→fail | 23,998 | 20,852 | -13% | 1 | 1 | 0% | 2,998 | 3,345 | +12% | 0 | 0 | — |
case-17 | pass→pass | 15,623 | 11,735 | -25% | 1 | 1 | 0% | 2,316 | 2,313 | -0% | 0 | 0 | — |
case-18 | fail→pass | 11,905 | 10,288 | -14% | 1 | 1 | 0% | 1,731 | 2,178 | +26% | 0 | 0 | — |
case-19 | pass→pass | 18,257 | 15,763 | -14% | 1 | 1 | 0% | 2,379 | 2,427 | +2% | 0 | 0 | — |
case-20 | fail→fail | 25,551 | 29,745 | +16% | 1 | 1 | 0% | 3,572 | 4,129 | +16% | 0 | 0 | — |
case-21 | fail→pass | 15,404 | 4,605 | -70% | 1 | 1 | 0% | 1,928 | 1,423 | -26% | 0 | 0 | — |
case-22 | fail→fail | 20,961 | 77,398 | +269% | 1 | 1 | 0% | 2,332 | 4,522 | +94% | 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 +36 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.