Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股结构变点检测/趋势拐点识别。当用户说"结构变点"、"structural break"、"拐点"、"趋势改变"、"什么时候变了"、"Chow检验"时触发。量化检测价格序列的结构性变化。支持formal和brief风格。
.claude/skills/aifinlab-a-share-structural-break/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 4% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [日期] python "$SCRIPTS/cn_stock_data.py" quote --code [CODE] python "$SCRIPTS/cn_stock_data.py" finance --code [CODE]
比较断点前后的均值/波动率/趋势斜率变化
关联断点时间与重大事件（政策/业绩/市场）
| 维度 | formal | brief | |------|--------|-------| | 断点检测 | 多方法+统计检验 | 最近断点 | | 前后对比 | 均值/波动率变化 | 趋势变了吗 | | 归因 | 事件关联分析 | 可能原因 | 默认风格：brief。
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 | 25,484 | 9,121 | -64% | 1 | 1 | 0% | 3,621 | 1,004 | -72% | 0 | 0 | — |
case-02 | fail→fail | 23,540 | 9,820 | -58% | 1 | 1 | 0% | 4,151 | 1,001 | -76% | 0 | 0 | — |
case-03 | fail→fail | 27,796 | 11,939 | -57% | 1 | 1 | 0% | 4,412 | 1,260 | -71% | 0 | 0 | — |
case-04 | fail→pass | 12,278 | 4,728 | -61% | 1 | 1 | 0% | 2,038 | 1,158 | -43% | 0 | 0 | — |
case-05 | fail→pass | 15,907 | 7,316 | -54% | 1 | 1 | 0% | 2,332 | 1,461 | -37% | 0 | 0 | — |
case-06 | pass→pass | 8,058 | 12,942 | +61% | 1 | 1 | 0% | 1,267 | 2,011 | +59% | 0 | 0 | — |
case-19 | fail→pass | 14,303 | 3,178 | -78% | 1 | 1 | 0% | 1,988 | 893 | -55% | 0 | 0 | — |
case-07 | pass→pass | 14,767 | 13,473 | -9% | 1 | 1 | 0% | 2,067 | 2,571 | +24% | 0 | 0 | — |
case-08 | pass→pass | 14,464 | 10,830 | -25% | 1 | 1 | 0% | 2,031 | 2,177 | +7% | 0 | 0 | — |
case-09 | fail→pass | 16,553 | 15,308 | -8% | 1 | 1 | 0% | 2,692 | 1,848 | -31% | 0 | 0 | — |
case-10 | pass→pass | 26,772 | 13,215 | -51% | 1 | 1 | 0% | 3,340 | 2,647 | -21% | 0 | 0 | — |
case-20 | pass→fail | 21,082 | 10,885 | -48% | 1 | 1 | 0% | 3,628 | 1,046 | -71% | 0 | 0 | — |
case-21 | pass→fail | 27,438 | 68,621 | +150% | 1 | 1 | 0% | 5,072 | 879 | -83% | 0 | 0 | — |
case-11 | fail→pass | 20,379 | 22,814 | +12% | 1 | 1 | 0% | 3,125 | 3,258 | +4% | 0 | 0 | — |
case-12 | pass→pass | 19,697 | 15,039 | -24% | 1 | 1 | 0% | 2,752 | 2,690 | -2% | 0 | 0 | — |
case-13 | fail→fail | 16,851 | 14,351 | -15% | 1 | 1 | 0% | 2,301 | 2,370 | +3% | 0 | 0 | — |
case-14 | pass→pass | 20,403 | 19,311 | -5% | 1 | 1 | 0% | 2,650 | 3,037 | +15% | 0 | 0 | — |
case-22 | pass→fail | 21,173 | 10,656 | -50% | 1 | 1 | 0% | 3,925 | 1,007 | -74% | 0 | 0 | — |
case-15 | fail→fail | 21,763 | 29,471 | +35% | 1 | 1 | 0% | 3,120 | 2,793 | -10% | 0 | 0 | — |
case-16 | fail→pass | 21,375 | 11,666 | -45% | 1 | 1 | 0% | 2,883 | 2,539 | -12% | 0 | 0 | — |
case-17 | pass→pass | 22,229 | 14,445 | -35% | 1 | 1 | 0% | 2,993 | 1,599 | -47% | 0 | 0 | — |
case-18 | fail→pass | 16,055 | 6,481 | -60% | 1 | 1 | 0% | 2,458 | 1,582 | -36% | 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, and 16 counted toward the lift figure. The other 6 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 +18 percentage points is the difference between those two pass rates over the 16 comparable cases. 5 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.