Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当需要检查业务留痕材料的完整性、可追溯性并输出缺失项与补件建议时使用本技能，适用于合规运营与质检场景。
.claude/skills/aifinlab-bank-t228-compliance-operations-traceability-check-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 33% | 0% |
本技能用于检查留痕材料是否满足可追溯与完整性要求，输出缺失项、异常项与补件建议，适合合规运营、内控与质检团队。
scripts/traceability_checker.py| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 25,592 | 25,582 | -0% | 1 | 1 | 0% | 3,930 | 4,183 | +6% | 0 | 0 | — |
case-02 | pass→pass | 6,556 | 15,950 | +143% | 1 | 1 | 0% | 830 | 2,417 | +191% | 0 | 0 | — |
case-03 | pass→fail | 15,049 | 14,942 | -1% | 1 | 1 | 0% | 2,215 | 2,487 | +12% | 0 | 0 | — |
case-04 | fail→pass | 19,730 | 13,831 | -30% | 1 | 1 | 0% | 2,529 | 2,345 | -7% | 0 | 0 | — |
case-05 | fail→pass | 28,251 | 21,107 | -25% | 1 | 1 | 0% | 3,470 | 3,647 | +5% | 0 | 0 | — |
case-06 | fail→pass | 22,857 | 15,412 | -33% | 1 | 1 | 0% | 3,819 | 2,424 | -37% | 0 | 0 | — |
case-07 | pass→pass | 19,851 | 18,532 | -7% | 1 | 1 | 0% | 2,637 | 3,003 | +14% | 0 | 0 | — |
case-08 | pass→pass | 19,955 | 22,701 | +14% | 1 | 1 | 0% | 2,963 | 3,330 | +12% | 0 | 0 | — |
case-09 | pass→pass | 19,120 | 18,139 | -5% | 1 | 1 | 0% | 2,605 | 2,813 | +8% | 0 | 0 | — |
case-10 | pass→pass | 20,860 | 22,167 | +6% | 1 | 1 | 0% | 2,550 | 3,154 | +24% | 0 | 0 | — |
case-11 | pass→pass | 21,170 | 20,454 | -3% | 1 | 1 | 0% | 3,062 | 3,006 | -2% | 0 | 0 | — |
case-12 | pass→pass | 24,872 | 25,011 | +1% | 1 | 1 | 0% | 3,555 | 3,687 | +4% | 0 | 0 | — |
case-13 | pass→pass | 19,568 | 18,446 | -6% | 1 | 1 | 0% | 2,390 | 3,393 | +42% | 0 | 0 | — |
case-14 | pass→pass | 17,321 | 19,206 | +11% | 1 | 1 | 0% | 2,342 | 2,871 | +23% | 0 | 0 | — |
case-15 | pass→pass | 18,015 | 36,299 | +101% | 1 | 1 | 0% | 2,652 | 3,114 | +17% | 0 | 0 | — |
case-16 | fail→pass | 18,124 | 22,282 | +23% | 1 | 1 | 0% | 2,348 | 3,120 | +33% | 0 | 0 | — |
case-17 | pass→pass | 20,042 | 17,625 | -12% | 1 | 1 | 0% | 2,474 | 2,991 | +21% | 0 | 0 | — |
case-18 | pass→pass | 15,117 | 46,205 | +206% | 1 | 1 | 0% | 1,949 | 2,975 | +53% | 0 | 0 | — |
case-19 | pass→pass | 26,890 | 21,231 | -21% | 1 | 1 | 0% | 2,610 | 3,237 | +24% | 0 | 0 | — |
case-20 | pass→pass | 25,603 | 24,003 | -6% | 1 | 1 | 0% | 3,077 | 3,610 | +17% | 0 | 0 | — |
case-21 | pass→pass | 19,843 | 26,417 | +33% | 1 | 1 | 0% | 3,002 | 3,465 | +15% | 0 | 0 | — |
case-22 | pass→pass | 34,748 | 19,020 | -45% | 1 | 1 | 0% | 2,651 | 3,067 | +16% | 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 +18 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.
Other measured skills in the registry, with their headline benchmark lift.