Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股财务异常/财务造假预警量化。当用户说"财务异常"、"financial forensic"、"造假"、"财务造假"、"Beneish"、"M-score"、"财务粉饰"时触发。量化检测财务报表异常信号。支持formal和brief风格。
.claude/skills/aifinlab-a-share-financial-forensic/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-15 | ✓→✗ | ▼ Worse | -82% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -11% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -3% | 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]
计算8个变量的加权得分（>-1.78为操纵嫌疑）
多维度财务异常打分
| 维度 | formal | brief | |------|--------|-------| | M-Score | 各变量明细 | 综合得分 | | 异常指标 | 全面检测结果 | 红旗数量 | | 风险等级 | 历史对比分析 | 高/中/低 | 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-08 | fail→pass | 23,805 | 3,785 | -84% | 1 | 1 | 0% | 4,292 | 1,194 | -72% | 0 | 0 | — |
case-01 | fail→fail | 25,487 | 19,423 | -24% | 1 | 1 | 0% | 4,245 | 1,457 | -66% | 0 | 0 | — |
case-02 | fail→fail | 49,360 | 9,169 | -81% | 1 | 1 | 0% | 7,859 | 960 | -88% | 0 | 0 | — |
case-03 | fail→fail | 22,524 | 7,694 | -66% | 1 | 1 | 0% | 4,056 | 875 | -78% | 0 | 0 | — |
case-04 | pass→pass | 14,263 | 10,588 | -26% | 1 | 1 | 0% | 2,211 | 1,973 | -11% | 0 | 0 | — |
case-05 | pass→pass | 19,782 | 17,103 | -14% | 1 | 1 | 0% | 2,997 | 2,922 | -3% | 0 | 0 | — |
case-06 | pass→pass | 22,151 | 17,998 | -19% | 1 | 1 | 0% | 2,796 | 3,070 | +10% | 0 | 0 | — |
case-07 | pass→pass | 20,541 | 23,021 | +12% | 1 | 1 | 0% | 2,976 | 3,992 | +34% | 0 | 0 | — |
case-09 | pass→pass | 14,758 | 9,680 | -34% | 1 | 1 | 0% | 2,158 | 1,975 | -8% | 0 | 0 | — |
case-10 | pass→pass | 19,145 | 14,505 | -24% | 1 | 1 | 0% | 2,788 | 2,589 | -7% | 0 | 0 | — |
case-11 | pass→pass | 15,657 | 14,934 | -5% | 1 | 1 | 0% | 2,478 | 2,876 | +16% | 0 | 0 | — |
case-12 | pass→pass | 17,925 | 16,708 | -7% | 1 | 1 | 0% | 2,701 | 2,901 | +7% | 0 | 0 | — |
case-13 | pass→pass | 14,180 | 11,349 | -20% | 1 | 1 | 0% | 2,148 | 2,250 | +5% | 0 | 0 | — |
case-14 | fail→fail | 16,368 | 8,519 | -48% | 1 | 1 | 0% | 2,467 | 852 | -65% | 0 | 0 | — |
case-15 | pass→fail | 39,920 | 7,852 | -80% | 1 | 1 | 0% | 5,625 | 1,001 | -82% | 0 | 0 | — |
case-16 | fail→fail | 8,797 | 8,710 | -1% | 1 | 1 | 0% | 1,458 | 1,895 | +30% | 0 | 0 | — |
case-17 | pass→pass | 6,287 | 7,935 | +26% | 1 | 1 | 0% | 1,095 | 1,636 | +49% | 0 | 0 | — |
case-18 | fail→pass | 24,971 | 14,995 | -40% | 1 | 1 | 0% | 3,363 | 3,188 | -5% | 0 | 0 | — |
case-19 | pass→pass | 19,981 | 20,118 | +1% | 1 | 1 | 0% | 2,902 | 3,236 | +12% | 0 | 0 | — |
case-20 | pass→pass | 8,764 | 6,154 | -30% | 1 | 1 | 0% | 1,626 | 1,628 | +0% | 0 | 0 | — |
case-21 | pass→pass | 20,274 | 15,363 | -24% | 1 | 1 | 0% | 2,881 | 3,238 | +12% | 0 | 0 | — |
case-22 | pass→pass | 21,797 | 21,157 | -3% | 1 | 1 | 0% | 3,153 | 4,504 | +43% | 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 17 counted toward the lift figure. The other 5 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 +5 percentage points is the difference between those two pass rates over the 17 comparable cases. 4 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | +13% |
Other measured skills in the registry, with their headline benchmark lift.