Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行交易银行与普惠场景下，围绕保函辅助形成结构化分析、判断和标准化输出时使用本技能。适合输出清晰的输入要求、处理步骤、结果结构和风险边界。
.claude/skills/aifinlab-bank-t247-transaction-banking-inclusive-finance-guarantee-support-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 63% | 0% |
本技能面向交易银行与普惠金融场景，既要兼顾准入效率和材料真实性，也要把结算、现金管理、供应链和贸易融资方案写到能落地的层面。 当前这支 skill 更偏向服务 交易银行和普惠团队，输出时要特别注意 兼顾准入效率、方案落地和风险边界。
本技能提供保函材料核验与输出模板脚本，用于快速生成保函核验报告。
scripts/guarantee_support_report.py用途：基于保函信息与材料清单输出核验结论与补件清单。
输入字段（JSON）：
guarantee_info: 保函类型、金额、期限、受益人contracts: 合同与履约材料margin_info: 保证金与授信占用materials: 材料清单rules: 规则与必备材料清单输出内容：
命令行示例：
python scripts/guarantee_support_report.py --input assets/guarantee_input.json --output outputs/guarantee_report.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | pass→pass | 21,871 | 25,136 | +15% | 1 | 1 | 0% | 2,942 | 4,338 | +47% | 0 | 0 | — |
case-12 | fail→fail | 18,532 | 24,496 | +32% | 1 | 1 | 0% | 2,560 | 4,630 | +81% | 0 | 0 | — |
case-01 | fail→fail | 25,344 | 64,515 | +155% | 1 | 1 | 0% | 3,589 | 6,160 | +72% | 0 | 0 | — |
case-02 | fail→fail | 20,268 | 22,381 | +10% | 1 | 1 | 0% | 2,721 | 4,219 | +55% | 0 | 0 | — |
case-03 | fail→pass | 16,874 | 18,650 | +11% | 1 | 1 | 0% | 2,210 | 3,552 | +61% | 0 | 0 | — |
case-04 | fail→fail | 50,031 | 21,095 | -58% | 1 | 1 | 0% | 2,670 | 4,137 | +55% | 0 | 0 | — |
case-05 | pass→pass | 19,283 | 23,721 | +23% | 1 | 1 | 0% | 2,781 | 4,261 | +53% | 0 | 0 | — |
case-06 | fail→fail | 21,308 | 21,056 | -1% | 1 | 1 | 0% | 2,718 | 3,776 | +39% | 0 | 0 | — |
case-07 | pass→pass | 15,323 | 24,008 | +57% | 1 | 1 | 0% | 2,122 | 4,230 | +99% | 0 | 0 | — |
case-08 | pass→pass | 14,912 | 17,137 | +15% | 1 | 1 | 0% | 1,985 | 3,403 | +71% | 0 | 0 | — |
case-09 | fail→fail | 23,325 | 20,857 | -11% | 1 | 1 | 0% | 3,136 | 3,878 | +24% | 0 | 0 | — |
case-10 | fail→pass | 18,514 | 18,334 | -1% | 1 | 1 | 0% | 2,413 | 3,548 | +47% | 0 | 0 | — |
case-13 | fail→pass | 20,027 | 20,475 | +2% | 1 | 1 | 0% | 2,816 | 3,910 | +39% | 0 | 0 | — |
case-14 | fail→fail | 17,268 | 28,646 | +66% | 1 | 1 | 0% | 2,415 | 5,197 | +115% | 0 | 0 | — |
case-15 | fail→pass | 21,639 | 24,337 | +12% | 1 | 1 | 0% | 2,769 | 4,426 | +60% | 0 | 0 | — |
case-16 | pass→pass | 20,453 | 25,706 | +26% | 1 | 1 | 0% | 2,702 | 4,620 | +71% | 0 | 0 | — |
case-17 | fail→pass | 16,293 | 21,034 | +29% | 1 | 1 | 0% | 2,361 | 3,856 | +63% | 0 | 0 | — |
case-18 | fail→pass | 19,238 | 17,964 | -7% | 1 | 1 | 0% | 2,552 | 3,527 | +38% | 0 | 0 | — |
case-19 | fail→fail | 15,663 | 24,884 | +59% | 1 | 1 | 0% | 2,182 | 4,536 | +108% | 0 | 0 | — |
case-20 | fail→pass | 15,105 | 21,664 | +43% | 1 | 1 | 0% | 2,031 | 3,966 | +95% | 0 | 0 | — |
case-21 | fail→pass | 20,003 | 23,803 | +19% | 1 | 1 | 0% | 2,578 | 4,235 | +64% | 0 | 0 | — |
case-22 | fail→pass | 15,112 | 17,231 | +14% | 1 | 1 | 0% | 2,166 | 3,452 | +59% | 0 | 0 | — |
case-23 | pass→pass | 19,672 | 25,541 | +30% | 1 | 1 | 0% | 2,693 | 4,707 | +75% | 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 +39 percentage points is the difference between those two pass rates over the 23 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.