Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股风险预警/ST预警/退市风险排查。当用户说"风险预警"、"有没有风险"、"会不会ST"、"退市风险"、"爆雷"、"暴雷"、"风险排查"、"XX安全吗"、"财务造假"、"审计意见"时触发。基于 cn-stock-data 财务数据，从盈利、现金流、负债、审计等维度排查个股风险信号。支持研报风格（formal）和快速排查风格（brief）。
.claude/skills/aifinlab-a-share-risk-alert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 37% | 0% |
对照 references/risk-alert-guide.md 中的10大风险信号逐项排查：
每项标注：✅ 正常 / ⚠️ 预警 / 🔴 高危
| 触发信号数 | 等级 | 标签 | |-----------|------|------| | 0-2 项 | 低风险 | 🟢 | | 3-4 项 | 中风险 | 🟡 | | 5-6 项 | 高风险 | 🔴 | | 7+ 项 | 极高风险 | ⛔ |
结合行业特殊性调整（见参考文档）
formal 模式（默认）：
brief 模式：
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 | 26,083 | 20,815 | -20% | 1 | 1 | 0% | 3,991 | 1,360 | -66% | 0 | 0 | — |
case-02 | fail→fail | 7,301 | 10,117 | +39% | 1 | 1 | 0% | 1,091 | 1,645 | +51% | 0 | 0 | — |
case-03 | fail→fail | 29,616 | 10,380 | -65% | 1 | 1 | 0% | 3,851 | 1,367 | -65% | 0 | 0 | — |
case-04 | pass→fail | 20,294 | 9,988 | -51% | 1 | 1 | 0% | 3,220 | 1,679 | -48% | 0 | 0 | — |
case-05 | fail→fail | 24,562 | 36,934 | +50% | 1 | 1 | 0% | 3,892 | 1,693 | -57% | 0 | 0 | — |
case-06 | fail→fail | 20,966 | 10,810 | -48% | 1 | 1 | 0% | 3,313 | 1,807 | -45% | 0 | 0 | — |
case-07 | fail→fail | 22,638 | 10,428 | -54% | 1 | 1 | 0% | 3,357 | 1,417 | -58% | 0 | 0 | — |
case-19 | fail→pass | 20,959 | 7,845 | -63% | 1 | 1 | 0% | 2,612 | 1,900 | -27% | 0 | 0 | — |
case-20 | pass→fail | 22,463 | 9,637 | -57% | 1 | 1 | 0% | 3,779 | 1,335 | -65% | 0 | 0 | — |
case-08 | fail→pass | 26,111 | 19,777 | -24% | 1 | 1 | 0% | 3,558 | 3,979 | +12% | 0 | 0 | — |
case-09 | pass→fail | 19,144 | 9,312 | -51% | 1 | 1 | 0% | 2,960 | 1,255 | -58% | 0 | 0 | — |
case-10 | fail→pass | 9,989 | 13,848 | +39% | 1 | 1 | 0% | 1,635 | 2,863 | +75% | 0 | 0 | — |
case-11 | fail→fail | 14,219 | 8,945 | -37% | 1 | 1 | 0% | 1,835 | 1,287 | -30% | 0 | 0 | — |
case-12 | fail→pass | 21,961 | 19,138 | -13% | 1 | 1 | 0% | 3,400 | 3,847 | +13% | 0 | 0 | — |
case-21 | pass→fail | 25,219 | 11,496 | -54% | 1 | 1 | 0% | 3,542 | 1,455 | -59% | 0 | 0 | — |
case-13 | pass→pass | 16,887 | 8,620 | -49% | 1 | 1 | 0% | 2,644 | 2,278 | -14% | 0 | 0 | — |
case-14 | fail→fail | 23,441 | 12,853 | -45% | 1 | 1 | 0% | 3,444 | 1,546 | -55% | 0 | 0 | — |
case-15 | fail→pass | 19,800 | 19,704 | -0% | 1 | 1 | 0% | 2,707 | 3,706 | +37% | 0 | 0 | — |
case-16 | fail→pass | 25,809 | 21,121 | -18% | 1 | 1 | 0% | 3,900 | 4,078 | +5% | 0 | 0 | — |
case-17 | pass→fail | 22,128 | 11,135 | -50% | 1 | 1 | 0% | 3,501 | 1,479 | -58% | 0 | 0 | — |
case-18 | pass→pass | 23,812 | 13,394 | -44% | 1 | 1 | 0% | 3,209 | 3,002 | -6% | 0 | 0 | — |
case-22 | pass→pass | 30,911 | 45,219 | +46% | 1 | 1 | 0% | 5,836 | 9,036 | +55% | 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 9 counted toward the lift figure. The other 13 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 9 comparable cases. 11 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 | +17% |
Other measured skills in the registry, with their headline benchmark lift.