Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行零售金融场景下，围绕资料核验进行完整性、一致性、真实性或规则符合性检查时使用本技能。适合输出核验结论、异常项清单、补件要求和升级复核建议。
.claude/skills/aifinlab-bank-t172-retail-finance-material-validation-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -19% | 0% |
在零售金融场景中对社保、公积金、收入证明等材料进行完整性、一致性与时效性核验，输出异常项与补件建议，支持资料核验与人工复核。
batch_id、time_window（建议包含 start_date/end_date）materials[]（建议字段 material_type、fields、source_date）id_number、employer_name、contribution_base、monthly_deposit、monthly_income、last_payment_month脚本入口：scripts/run_skill.py，调用 shared/retail_service_skill_engine.py 的 t172 场景。
bashpython scripts/run_skill.py --input input.json --format markdown python scripts/run_skill.py --input input.json --format 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 | 16,884 | 32,575 | +93% | 1 | 1 | 0% | 2,557 | 4,419 | +73% | 0 | 0 | — |
case-02 | fail→pass | 20,220 | 17,445 | -14% | 1 | 1 | 0% | 3,029 | 3,412 | +13% | 0 | 0 | — |
case-03 | fail→pass | 19,712 | 34,238 | +74% | 1 | 1 | 0% | 2,897 | 4,113 | +42% | 0 | 0 | — |
case-04 | pass→pass | 11,565 | 14,771 | +28% | 1 | 1 | 0% | 1,373 | 3,043 | +122% | 0 | 0 | — |
case-09 | pass→pass | 16,405 | 19,259 | +17% | 1 | 1 | 0% | 2,088 | 3,277 | +57% | 0 | 0 | — |
case-05 | pass→pass | 6,762 | 8,785 | +30% | 1 | 1 | 0% | 924 | 1,961 | +112% | 0 | 0 | — |
case-06 | pass→pass | 11,420 | 14,035 | +23% | 1 | 1 | 0% | 1,521 | 2,433 | +60% | 0 | 0 | — |
case-07 | pass→pass | 22,714 | 16,652 | -27% | 1 | 1 | 0% | 2,835 | 3,368 | +19% | 0 | 0 | — |
case-08 | pass→pass | 15,855 | 18,438 | +16% | 1 | 1 | 0% | 2,563 | 3,516 | +37% | 0 | 0 | — |
case-10 | fail→pass | 14,656 | 17,989 | +23% | 1 | 1 | 0% | 2,192 | 2,948 | +34% | 0 | 0 | — |
case-11 | pass→pass | 20,623 | 23,053 | +12% | 1 | 1 | 0% | 2,720 | 3,552 | +31% | 0 | 0 | — |
case-12 | fail→pass | 8,592 | 2,660 | -69% | 1 | 1 | 0% | 1,410 | 1,111 | -21% | 0 | 0 | — |
case-13 | fail→pass | 10,318 | 6,182 | -40% | 1 | 1 | 0% | 1,760 | 1,423 | -19% | 0 | 0 | — |
case-18 | pass→pass | 9,063 | 14,993 | +65% | 1 | 1 | 0% | 1,240 | 2,739 | +121% | 0 | 0 | — |
case-14 | pass→pass | 15,046 | 18,507 | +23% | 1 | 1 | 0% | 2,071 | 3,020 | +46% | 0 | 0 | — |
case-15 | pass→pass | 17,179 | 17,186 | +0% | 1 | 1 | 0% | 2,200 | 3,034 | +38% | 0 | 0 | — |
case-16 | pass→pass | 13,314 | 16,257 | +22% | 1 | 1 | 0% | 1,871 | 3,028 | +62% | 0 | 0 | — |
case-17 | pass→pass | 15,808 | 18,722 | +18% | 1 | 1 | 0% | 2,304 | 3,482 | +51% | 0 | 0 | — |
case-19 | pass→pass | 20,107 | 20,526 | +2% | 1 | 1 | 0% | 3,433 | 3,924 | +14% | 0 | 0 | — |
case-20 | pass→pass | 14,749 | 18,450 | +25% | 1 | 1 | 0% | 2,211 | 3,211 | +45% | 0 | 0 | — |
case-21 | pass→pass | 23,102 | 25,818 | +12% | 1 | 1 | 0% | 2,959 | 3,997 | +35% | 0 | 0 | — |
case-22 | pass→pass | 22,262 | 21,428 | -4% | 1 | 1 | 0% | 2,809 | 3,624 | +29% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.