Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当用户需要在银行对公金融场景下，为授信项目准备审批会材料、提炼支持与反对理由、识别核心争议点并形成待拍板事项时使用本技能。适合输出审批会一页纸、争议矩阵、有条件通过条款和会后推进动作。
.claude/skills/aifinlab-bank-t124-corporate-finance-creditdue-diligence-credit-committee-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 96% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 144% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 93% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 156% | 0% |
这个 skill 面向授信审批会场景使用。它不是最终批复工具，而是把客户经理、审查岗、风控、授信秘书等多方材料统一成“审批会可直接讨论和拍板”的结构化包。
它优先解决六类问题：
建议尽量提供：
详细字段见 input-schema.md。
默认输出至少包含：
scripts/corporate_credit_committee_packet.py：生成审批会材料包scripts/run_skill.py：脚本入口示例调用：
bashpython scripts/run_skill.py --input assets/example-input.json --output committee_packet.md --format markdown
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→fail | 19,059 | 22,224 | +17% | 1 | 1 | 0% | 2,741 | 5,357 | +95% | 0 | 0 | — |
case-01 | fail→fail | 25,339 | 37,339 | +47% | 1 | 1 | 0% | 3,997 | 6,032 | +51% | 0 | 0 | — |
case-02 | fail→fail | 31,844 | 34,663 | +9% | 1 | 1 | 0% | 4,701 | 7,067 | +50% | 0 | 0 | — |
case-03 | fail→pass | 44,952 | 41,796 | -7% | 1 | 1 | 0% | 4,269 | 8,380 | +96% | 0 | 0 | — |
case-04 | fail→pass | 15,841 | 25,416 | +60% | 1 | 1 | 0% | 2,373 | 5,789 | +144% | 0 | 0 | — |
case-05 | pass→pass | 11,008 | 13,324 | +21% | 1 | 1 | 0% | 1,503 | 3,735 | +149% | 0 | 0 | — |
case-06 | fail→pass | 19,552 | 34,746 | +78% | 1 | 1 | 0% | 2,849 | 5,505 | +93% | 0 | 0 | — |
case-07 | fail→pass | 20,831 | 23,943 | +15% | 1 | 1 | 0% | 3,072 | 5,675 | +85% | 0 | 0 | — |
case-08 | pass→pass | 18,018 | 28,387 | +58% | 1 | 1 | 0% | 2,537 | 6,087 | +140% | 0 | 0 | — |
case-09 | pass→pass | 20,619 | 40,078 | +94% | 1 | 1 | 0% | 2,837 | 7,732 | +173% | 0 | 0 | — |
case-10 | fail→pass | 19,329 | 38,900 | +101% | 1 | 1 | 0% | 2,743 | 7,025 | +156% | 0 | 0 | — |
case-11 | fail→pass | 32,588 | 34,401 | +6% | 1 | 1 | 0% | 3,857 | 7,205 | +87% | 0 | 0 | — |
case-12 | pass→pass | 19,707 | 21,118 | +7% | 1 | 1 | 0% | 2,751 | 4,977 | +81% | 0 | 0 | — |
case-13 | pass→pass | 27,031 | 34,649 | +28% | 1 | 1 | 0% | 3,512 | 6,524 | +86% | 0 | 0 | — |
case-14 | pass→pass | 21,095 | 28,646 | +36% | 1 | 1 | 0% | 2,648 | 5,461 | +106% | 0 | 0 | — |
case-15 | pass→pass | 27,153 | 32,186 | +19% | 1 | 1 | 0% | 3,668 | 6,746 | +84% | 0 | 0 | — |
case-17 | pass→pass | 21,995 | 33,543 | +53% | 1 | 1 | 0% | 3,323 | 6,821 | +105% | 0 | 0 | — |
case-18 | fail→pass | 26,698 | 38,508 | +44% | 1 | 1 | 0% | 3,330 | 6,972 | +109% | 0 | 0 | — |
case-19 | pass→pass | 18,792 | 16,302 | -13% | 1 | 1 | 0% | 2,560 | 4,328 | +69% | 0 | 0 | — |
case-20 | fail→pass | 29,096 | 36,821 | +27% | 1 | 1 | 0% | 3,925 | 6,373 | +62% | 0 | 0 | — |
case-21 | pass→pass | 21,207 | 22,216 | +5% | 1 | 1 | 0% | 2,407 | 4,944 | +105% | 0 | 0 | — |
case-22 | pass→pass | 18,262 | 28,516 | +56% | 1 | 1 | 0% | 2,523 | 5,960 | +136% | 0 | 0 | — |
case-23 | pass→pass | 21,349 | 41,326 | +94% | 1 | 1 | 0% | 2,730 | 6,959 | +155% | 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 +35 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is 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.