Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要对银行对公授信场景中的授信申报材料进行完整性核验、缺件识别、有效性检查、版本一致性检查、签章核验、补件清单生成、材料审核意见撰写时，使用此技能。 适用于授信材料初审、预审、送审前复核、补件管理、材料台账整理、审查清单核对等场景。 当用户要求输出材料缺失项、材料过期项、材料矛盾项、待补充说明项、审核结论、补件通知时，应优先触发本技能。
.claude/skills/aifinlab-credit-material-completeness-check/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 112% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 179% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-24 | ✗→✓ | ▲ Improved | 98% | 0% |
授信材料完整性核验助手
本技能用于银行对公金融授信场景下的材料审核工作，重点解决“材料是否齐全、是否有效、是否一致、是否满足送审要求”四类问题。
本技能不直接替代授信审批结论，也不对客户授信是否准入作最终判断；其主要职责是完成材料层面的审核、核对、归类、缺口识别与补件建议输出。
理想输入包括但不限于：
若用户未提供完整清单，也应先基于当前信息输出：
先识别以下核心信息：
若上述信息缺失，应在输出中单列“基础信息待补充”。
将用户提供的材料整理成标准台账，至少包含：
至少按以下类别进行检查：
重点核对以下字段在不同材料间是否一致：
如存在不一致，应标注为：
重点检查：
最终至少输出以下内容：
输出时建议采用以下结构：
若用户只给了一份材料清单，没有正文：
若用户只给了一堆材料正文，没有材料目录：
若用户要求写补件通知：
若用户要求写审核意见：
详细规则见：
references/material_checklist.mdreferences/review_rules.mdreferences/red_flags.mdreferences/output_schema.mdreferences/report_template.mdassets/templates/material_register_template.mdassets/templates/supplement_request_template.mdassets/templates/review_opinion_template.mdscripts/material_register_builder.pyscripts/document_consistency_checker.pyscripts/review_report_renderer.py| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 31,744 | 59,782 | +88% | 1 | 1 | 0% | 4,048 | 7,471 | +85% | 0 | 0 | — |
case-02 | fail→fail | 26,373 | 36,034 | +37% | 1 | 1 | 0% | 3,698 | 7,014 | +90% | 0 | 0 | — |
case-03 | fail→fail | 25,834 | 31,850 | +23% | 1 | 1 | 0% | 3,845 | 6,787 | +77% | 0 | 0 | — |
case-04 | fail→pass | 19,882 | 23,102 | +16% | 1 | 1 | 0% | 2,478 | 5,262 | +112% | 0 | 0 | — |
case-05 | fail→fail | 28,517 | 33,148 | +16% | 1 | 1 | 0% | 3,963 | 6,221 | +57% | 0 | 0 | — |
case-06 | pass→pass | 18,999 | 19,112 | +1% | 1 | 1 | 0% | 2,738 | 4,856 | +77% | 0 | 0 | — |
case-07 | pass→pass | 19,341 | 19,559 | +1% | 1 | 1 | 0% | 3,064 | 5,234 | +71% | 0 | 0 | — |
case-08 | fail→fail | 16,614 | 22,126 | +33% | 1 | 1 | 0% | 2,564 | 5,723 | +123% | 0 | 0 | — |
case-09 | pass→pass | 10,448 | 16,692 | +60% | 1 | 1 | 0% | 1,817 | 4,824 | +165% | 0 | 0 | — |
case-10 | fail→pass | 18,661 | 19,713 | +6% | 1 | 1 | 0% | 2,563 | 4,420 | +72% | 0 | 0 | — |
case-11 | pass→pass | 19,844 | 19,627 | -1% | 1 | 1 | 0% | 2,679 | 4,680 | +75% | 0 | 0 | — |
case-12 | pass→pass | 13,098 | 13,752 | +5% | 1 | 1 | 0% | 1,938 | 3,861 | +99% | 0 | 0 | — |
case-13 | pass→pass | 20,935 | 15,649 | -25% | 1 | 1 | 0% | 2,645 | 4,600 | +74% | 0 | 0 | — |
case-19 | pass→pass | 17,699 | 20,190 | +14% | 1 | 1 | 0% | 2,504 | 4,884 | +95% | 0 | 0 | — |
case-14 | pass→pass | 13,965 | 17,183 | +23% | 1 | 1 | 0% | 2,088 | 4,348 | +108% | 0 | 0 | — |
case-15 | pass→pass | 8,869 | 17,587 | +98% | 1 | 1 | 0% | 1,562 | 4,395 | +181% | 0 | 0 | — |
case-16 | fail→fail | 17,430 | 20,182 | +16% | 1 | 1 | 0% | 2,469 | 4,914 | +99% | 0 | 0 | — |
case-17 | fail→pass | 24,534 | 26,248 | +7% | 1 | 1 | 0% | 2,271 | 6,336 | +179% | 0 | 0 | — |
case-18 | fail→pass | 19,626 | 30,995 | +58% | 1 | 1 | 0% | 2,653 | 5,058 | +91% | 0 | 0 | — |
case-20 | pass→pass | 15,937 | 20,171 | +27% | 1 | 1 | 0% | 2,298 | 4,794 | +109% | 0 | 0 | — |
case-21 | pass→pass | 10,121 | 17,982 | +78% | 1 | 1 | 0% | 1,743 | 4,843 | +178% | 0 | 0 | — |
case-22 | pass→pass | 11,846 | 20,286 | +71% | 1 | 1 | 0% | 1,816 | 4,766 | +162% | 0 | 0 | — |
case-23 | pass→pass | 13,347 | 16,385 | +23% | 1 | 1 | 0% | 1,860 | 4,368 | +135% | 0 | 0 | — |
case-24 | fail→pass | 28,454 | 35,785 | +26% | 1 | 1 | 0% | 3,775 | 7,475 | +98% | 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. 24 cases were attempted. The headline lift of +21 percentage points is the difference between those two pass rates over the 24 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.