Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →开户审核助手。专注于客户开户资料审核，包括身份验证、风险测评、适当性匹配、资料完整性检查，帮助确保开户合规。 **触发场景**： - 用户需要审核客户开户资料 - 用户说"开户审核"、"客户审核"、"资料审核" - 需要身份验证、风险测评、适当性匹配 - 需要资料完整性检查、合规审核 **关键词**："开户"、"审核"、"客户"、"身份验证"、"风险测评"、"适当性"、"资料审核"
.claude/skills/aifinlab-account-opening-review-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-17 | ✓→✗ | ▼ Worse | 20% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 4% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 12% | 0% |
你是一名经验丰富的合规专员，擅长审核客户开户资料，帮助确保开户流程符合监管要求和公司合规标准。
| 要素 | 内容 | 要求 | |------|------|------| | 身份验证 | 身份证、人脸识别 | 真实有效、人证一致 | | 资料完整性 | 必填项、附件 | 完整无缺失 | | 风险测评 | 风险等级评估 | 真实有效、在有效期 | | 适当性匹配 | 产品与风险匹配 | 匹配或超配警示 | | 黑名单检查 | 禁入人员、失信人员 | 无禁入记录 |
markdown# 【开户审核报告】 ## 客户信息 - 姓名：[姓名] - 身份证号：[号码] - 联系方式：[电话] - 申请时间：[时间] ## 审核项目 ### 1. 身份验证 - 身份证：[有效/无效/过期] - 人脸识别：[通过/未通过] - 银行卡：[验证通过/未通过] - 结果：[通过/需复核/不通过] ### 2. 资料完整性 - 基本信息：[完整/缺失] - 职业信息：[完整/缺失] - 投资信息：[完整/缺失] - 结果：[通过/需补充] ### 3. 风险测评 - 测评时间：[时间] - 风险等级：[C1-C5] - 有效期：[是否有效] - 结果：[通过/需重测] ### 4. 适当性匹配 - 申请产品：[产品名称] - 产品风险：[R1-R5] - 客户风险：[C1-C5] - 匹配结果：[匹配/超配警示] ### 5. 黑名单检查 - 禁入人员：[是/否] - 失信人员：[是/否] - 反洗钱：[正常/可疑] - 结果：[通过/不通过] ## 审核结论 - [ ] 通过，可开户 - [ ] 需补充资料 - [ ] 需人工复核 - [ ] 不通过，拒绝开户 ## 补充/复核事项 - [事项 1] - [事项 2] --- 审核人：[姓名] 审核时间：[时间]
| 结论 | 标准 | |------|------| | 通过 | 所有项目通过 | | 需补充资料 | 资料不完整 | | 需人工复核 | 有疑点需人工判断 | | 不通过 | 身份虚假/黑名单/禁入 |
输出前自查：
python# 调用 skill result = run_skill({ "param1": "value1", "param2": "value2" })
bashpython scripts/run_skill.py --input data.json
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 13,915 | 10,155 | -27% | 1 | 1 | 0% | 2,437 | 2,849 | +17% | 0 | 0 | — |
case-02 | pass→pass | 20,436 | 10,297 | -50% | 1 | 1 | 0% | 3,143 | 3,266 | +4% | 0 | 0 | — |
case-03 | pass→pass | 20,209 | 10,738 | -47% | 1 | 1 | 0% | 3,006 | 3,375 | +12% | 0 | 0 | — |
case-04 | pass→pass | 16,821 | 8,409 | -50% | 1 | 1 | 0% | 2,678 | 2,946 | +10% | 0 | 0 | — |
case-05 | pass→pass | 13,874 | 10,133 | -27% | 1 | 1 | 0% | 2,379 | 3,160 | +33% | 0 | 0 | — |
case-06 | pass→pass | 16,864 | 12,990 | -23% | 1 | 1 | 0% | 2,902 | 3,085 | +6% | 0 | 0 | — |
case-07 | pass→pass | 11,016 | 8,089 | -27% | 1 | 1 | 0% | 1,978 | 2,784 | +41% | 0 | 0 | — |
case-08 | pass→pass | 14,009 | 10,634 | -24% | 1 | 1 | 0% | 2,057 | 2,791 | +36% | 0 | 0 | — |
case-09 | fail→pass | 14,143 | 10,539 | -25% | 1 | 1 | 0% | 2,379 | 3,363 | +41% | 0 | 0 | — |
case-10 | pass→pass | 17,391 | 9,393 | -46% | 1 | 1 | 0% | 2,501 | 2,991 | +20% | 0 | 0 | — |
case-11 | fail→fail | 14,860 | 9,576 | -36% | 1 | 1 | 0% | 2,148 | 3,063 | +43% | 0 | 0 | — |
case-12 | pass→pass | 14,312 | 8,371 | -42% | 1 | 1 | 0% | 2,349 | 2,673 | +14% | 0 | 0 | — |
case-13 | pass→pass | 15,971 | 6,989 | -56% | 1 | 1 | 0% | 2,140 | 2,663 | +24% | 0 | 0 | — |
case-14 | pass→pass | 14,597 | 11,798 | -19% | 1 | 1 | 0% | 2,490 | 2,905 | +17% | 0 | 0 | — |
case-15 | pass→pass | 12,280 | 7,939 | -35% | 1 | 1 | 0% | 2,028 | 2,682 | +32% | 0 | 0 | — |
case-16 | pass→pass | 17,682 | 10,408 | -41% | 1 | 1 | 0% | 2,855 | 3,103 | +9% | 0 | 0 | — |
case-17 | pass→fail | 15,430 | 11,392 | -26% | 1 | 1 | 0% | 2,320 | 2,778 | +20% | 0 | 0 | — |
case-18 | pass→pass | 13,613 | 8,381 | -38% | 1 | 1 | 0% | 2,266 | 2,791 | +23% | 0 | 0 | — |
case-19 | pass→pass | 15,713 | 7,828 | -50% | 1 | 1 | 0% | 2,301 | 2,754 | +20% | 0 | 0 | — |
case-20 | pass→pass | 23,553 | 21,978 | -7% | 1 | 1 | 0% | 2,851 | 4,168 | +46% | 0 | 0 | — |
case-21 | pass→pass | 24,022 | 16,945 | -29% | 1 | 1 | 0% | 3,944 | 3,774 | -4% | 0 | 0 | — |
case-22 | pass→pass | 19,780 | 26,390 | +33% | 1 | 1 | 0% | 2,625 | 4,407 | +68% | 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 0 percentage points is the difference between those two pass rates over the 22 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.