Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在交易银行与普惠金融场景下整理授信摘要、准入纪要、日报周报或催办跟踪时使用本技能，适合产出结构化摘要、补件清单、责任分工与后续动作。
.claude/skills/aifinlab-bank-t250-transaction-banking-inclusive-finance-credit-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -8% | 0% |
本技能用于交易银行与普惠金融授信场景的材料整合与摘要输出，目标是让客户经理、产品经理与运营支持在保证信息真实与风控边界的前提下，快速形成可执行的授信摘要、补件清单和后续推进动作。
本技能提供一个脚本用于结构化整理授信摘要与补件清单，适用于重复输出场景。
scripts/credit_summary_builder.py用途：把输入材料（JSON）转换为结构化授信摘要与补件清单（Markdown/JSON）。
输入字段（JSON）：
client_profile: 客户主体信息business_flow: 交易链路与资金流financing_request: 融资诉求与用途materials: 税票/流水/合同/票据/会议纪要等材料清单risks: 已识别风险点follow_up: 已有跟进事项output_type: summary/daily/weekly/reminder输出内容：
命令行示例：
python scripts/credit_summary_builder.py --input assets/credit_input.json --format markdown --output outputs/credit_summary.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→pass | 6,880 | 39,819 | +479% | 1 | 1 | 0% | 1,052 | 2,660 | +153% | 0 | 0 | — |
case-04 | pass→pass | 13,349 | 18,041 | +35% | 1 | 1 | 0% | 2,047 | 3,781 | +85% | 0 | 0 | — |
case-05 | pass→pass | 12,762 | 47,921 | +275% | 1 | 1 | 0% | 2,064 | 3,764 | +82% | 0 | 0 | — |
case-02 | pass→pass | 36,693 | 12,082 | -67% | 1 | 1 | 0% | 773 | 2,798 | +262% | 0 | 0 | — |
case-01 | fail→pass | 40,144 | 48,750 | +21% | 1 | 1 | 0% | 1,667 | 3,047 | +83% | 0 | 0 | — |
case-06 | pass→pass | 19,321 | 17,101 | -11% | 1 | 1 | 0% | 2,696 | 3,771 | +40% | 0 | 0 | — |
case-07 | fail→pass | 17,681 | 14,990 | -15% | 1 | 1 | 0% | 2,667 | 3,383 | +27% | 0 | 0 | — |
case-08 | fail→pass | 8,366 | 3,832 | -54% | 1 | 1 | 0% | 1,273 | 1,665 | +31% | 0 | 0 | — |
case-09 | fail→pass | 8,614 | 3,817 | -56% | 1 | 1 | 0% | 1,566 | 1,837 | +17% | 0 | 0 | — |
case-18 | fail→pass | 9,545 | 2,576 | -73% | 1 | 1 | 0% | 1,643 | 1,507 | -8% | 0 | 0 | — |
case-10 | fail→pass | 9,829 | 31,710 | +223% | 1 | 1 | 0% | 1,674 | 1,406 | -16% | 0 | 0 | — |
case-11 | fail→pass | 5,277 | 1,719 | -67% | 1 | 1 | 0% | 918 | 1,378 | +50% | 0 | 0 | — |
case-12 | pass→pass | 18,256 | 23,866 | +31% | 1 | 1 | 0% | 2,733 | 4,262 | +56% | 0 | 0 | — |
case-13 | pass→pass | 9,795 | 7,985 | -18% | 1 | 1 | 0% | 1,494 | 2,330 | +56% | 0 | 0 | — |
case-14 | fail→pass | 17,748 | 11,609 | -35% | 1 | 1 | 0% | 2,500 | 3,217 | +29% | 0 | 0 | — |
case-15 | pass→pass | 14,266 | 9,377 | -34% | 1 | 1 | 0% | 2,324 | 2,699 | +16% | 0 | 0 | — |
case-16 | fail→pass | 7,982 | 2,434 | -70% | 1 | 1 | 0% | 1,414 | 1,528 | +8% | 0 | 0 | — |
case-17 | fail→pass | 9,000 | 2,191 | -76% | 1 | 1 | 0% | 1,435 | 1,534 | +7% | 0 | 0 | — |
case-19 | pass→pass | 9,929 | 5,489 | -45% | 1 | 1 | 0% | 1,616 | 1,957 | +21% | 0 | 0 | — |
case-20 | pass→pass | 15,331 | 13,301 | -13% | 1 | 1 | 0% | 2,394 | 3,292 | +38% | 0 | 0 | — |
case-21 | pass→pass | 6,572 | 2,764 | -58% | 1 | 1 | 0% | 1,056 | 1,548 | +47% | 0 | 0 | — |
case-22 | pass→pass | 12,152 | 12,766 | +5% | 1 | 1 | 0% | 1,698 | 2,987 | +76% | 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 +45 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.