Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行合规运营场景下，对质检抽样进行完整性、一致性、真实性与规则符合性核验，并形成异常项、补件要求和升级复核建议时使用本技能。
.claude/skills/aifinlab-bank-t230-compliance-operations-quality-review-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 13% | 0% |
本技能用于银行合规运营质检场景，把制度口径、质检规则和留痕要求转成可执行的核验结论。输出强调可追溯、可复核和可落地，便于合规、运营和质检团队直接推进整改。
scripts/quality_review_checker.py：用于对抽样记录进行字段完整性、一致性与规则命中核验，输出异常项与补件清单。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 20,169 | 24,524 | +22% | 1 | 1 | 0% | 2,897 | 3,850 | +33% | 0 | 0 | — |
case-02 | fail→fail | 18,107 | 38,015 | +110% | 1 | 1 | 0% | 2,744 | 5,581 | +103% | 0 | 0 | — |
case-03 | fail→fail | 18,892 | 24,839 | +31% | 1 | 1 | 0% | 2,412 | 3,992 | +66% | 0 | 0 | — |
case-04 | fail→fail | 9,500 | 14,375 | +51% | 1 | 1 | 0% | 1,400 | 3,004 | +115% | 0 | 0 | — |
case-05 | pass→pass | 11,215 | 10,604 | -5% | 1 | 1 | 0% | 1,533 | 2,000 | +30% | 0 | 0 | — |
case-06 | fail→pass | 20,990 | 16,099 | -23% | 1 | 1 | 0% | 2,943 | 2,934 | -0% | 0 | 0 | — |
case-21 | pass→fail | 13,472 | 16,041 | +19% | 1 | 1 | 0% | 2,726 | 3,548 | +30% | 0 | 0 | — |
case-22 | pass→pass | 18,881 | 22,864 | +21% | 1 | 1 | 0% | 2,555 | 3,906 | +53% | 0 | 0 | — |
case-07 | fail→pass | 17,880 | 13,717 | -23% | 1 | 1 | 0% | 2,704 | 2,911 | +8% | 0 | 0 | — |
case-08 | fail→pass | 16,355 | 14,415 | -12% | 1 | 1 | 0% | 1,842 | 2,733 | +48% | 0 | 0 | — |
case-09 | pass→pass | 15,635 | 17,531 | +12% | 1 | 1 | 0% | 1,829 | 2,757 | +51% | 0 | 0 | — |
case-10 | fail→pass | 16,490 | 16,297 | -1% | 1 | 1 | 0% | 2,431 | 2,757 | +13% | 0 | 0 | — |
case-11 | fail→fail | 24,536 | 22,113 | -10% | 1 | 1 | 0% | 3,116 | 3,878 | +24% | 0 | 0 | — |
case-12 | fail→pass | 15,860 | 18,563 | +17% | 1 | 1 | 0% | 1,981 | 2,911 | +47% | 0 | 0 | — |
case-13 | pass→pass | 13,495 | 17,611 | +31% | 1 | 1 | 0% | 1,881 | 2,850 | +52% | 0 | 0 | — |
case-14 | fail→pass | 9,281 | 11,084 | +19% | 1 | 1 | 0% | 1,416 | 2,285 | +61% | 0 | 0 | — |
case-15 | fail→pass | 15,065 | 14,793 | -2% | 1 | 1 | 0% | 1,980 | 2,531 | +28% | 0 | 0 | — |
case-16 | pass→pass | 14,718 | 16,526 | +12% | 1 | 1 | 0% | 1,936 | 2,849 | +47% | 0 | 0 | — |
case-17 | pass→pass | 15,036 | 17,135 | +14% | 1 | 1 | 0% | 1,824 | 2,960 | +62% | 0 | 0 | — |
case-18 | fail→pass | 9,461 | 7,472 | -21% | 1 | 1 | 0% | 1,401 | 1,724 | +23% | 0 | 0 | — |
case-19 | pass→pass | 16,156 | 20,142 | +25% | 1 | 1 | 0% | 2,508 | 3,418 | +36% | 0 | 0 | — |
case-20 | pass→fail | 24,708 | 24,704 | -0% | 1 | 1 | 0% | 2,879 | 4,573 | +59% | 0 | 0 | — |
case-23 | pass→pass | 10,429 | 9,677 | -7% | 1 | 1 | 0% | 1,696 | 2,319 | +37% | 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. 23 cases were attempted. The headline lift of +30 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 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.