← Back to Skills Marketplace
gechengling

Insurance Anti-Fraud Expert

by gechengling · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ✓ Security Clean
102
Downloads
1
Stars
0
Active Installs
3
Versions
Install in OpenClaw
/install insurance-anti-fraud
Description
基于国家监管办法,提供保险理赔和核保欺诈识别、风险评分、调查流程及防范建议,助力风控合规。
README (SKILL.md)

\r \r

Insurance Anti-Fraud Expert / 保险反欺诈分析专家\r

\r

English: AI-powered insurance anti-fraud analysis expert — the definitive skill for detecting and preventing insurance fraud in the Chinese insurance market. Covers the complete CBIRC Anti-Insurance Fraud Measures (2024) framework, claim fraud identification engine (10 detection features), underwriting risk control system, fraud investigation workflows, and big data anti-fraud technology applications. Built for insurance company claims departments, risk control teams, and compliance teams.\r \r 中文: 保险反欺诈分析专家——中国《反保险欺诈工作办法》合规垂直Skill。覆盖四位一体反欺诈工作体系、理赔欺诈识别引擎(10大欺诈特征)、核保风控体系、欺诈案件调查流程(SOP)、大数据反欺诈技术应用。适用:保险公司理赔部、风控部、合规部、反欺诈调查员、保险公司核保岗。\r \r ---\r \r

Trigger Keywords / 触发关键词\r

\r English: anti-fraud, insurance fraud, claims fraud, fraud detection, risk control, underwriting risk, fraud investigation, money laundering, CBIRC compliance, claim review, insurance crime, fraud prevention, China insurance\r \r 中文触发词(优先): 反保险欺诈、保险欺诈识别、理赔欺诈、反欺诈调查、理赔风控、核保风控、道德风险、保险诈骗、欺诈特征、异常理赔、保险黑产、虚假投保、带病投保、理赔调查、欺诈渗漏、反欺诈模型\r \r ---\r \r

Core Capabilities / 核心能力\r

\r

1. Regulatory Framework: Anti-Insurance Fraud Measures / 监管框架:《反保险欺诈工作办法》全解\r

\r 法规来源:国家金融监督管理总局,2024年7月22日发布实施\r \r

四位一体反欺诈工作体系\r

\r

┌─────────────────────────────────────────────────┐\r
│                   四位一体体系                        │\r
├─────────────────────────────────────────────────┤\r
│  ① 行业反欺诈:银保信/精算师协会/保险行业协会协同    │\r
│  ② 公司主体责任:各保险公司反欺诈组织架构           │\r
│  ③ 公安司法机关:涉嫌犯罪的移送公安                │\r
│  ④ 监管部门:日常监管+专项检查+处罚                │\r
└─────────────────────────────────────────────────┘\r
```\r
\r
#### 欺诈类型分类\r
\r
| 类型 | 定义 | 典型场景 | 涉案险种 |\r
|------|------|---------|---------|\r
| **投保欺诈** | 故意不告知/虚假告知 | 带病投保、职业虚报 | 健康险、意外险 |\r
| **理赔欺诈** | 虚假索赔/夸大损失 | 伪造事故、虚构身份 | 车险、意外险、健康险 |\r
| **团伙欺诈** | 有组织犯罪团伙 | 专业碰瓷、产业链欺诈 | 医疗险、意外险 |\r
| **内部欺诈** | 内部人员参与 | 内外勾结、虚假理赔 | 各险种 |\r
| **洗钱型欺诈** | 借保险洗白资金 | 趸交高额寿险退保 | 寿险 |\r
\r
### 2. Claims Fraud Identification Engine / 理赔欺诈识别引擎\r
\r
#### 健康险理赔欺诈十大特征\r
\r
| 特征序号 | 欺诈特征 | 识别信号 |\r
|---------|---------|---------|\r
| 1 | **短时间内集中投保** | 等待期刚过即集中大量理赔 |\r
| 2 | **带病投保隐瞒病史** | 健康告知全为"否",理赔时发现既往症 |\r
| 3 | **虚假医院就诊** | 病历签名笔迹不一致、无实际治疗记录 |\r
| 4 | **过度医疗/过度检查** | 费用远超同类疾病平均水平的3倍以上 |\r
| 5 | **团伙作案特征** | 同一天、同一家医院、同一病种的群体报案 |\r
| 6 | **非正常高保额** | 收入与投保金额严重不符(意外险高保额) |\r
| 7 | **频繁小额理赔** | 利用免赔额规则反复小额索赔 |\r
| 8 | **跨公司重复索赔** | 利用信息不对称在多家公司重复索赔 |\r
| 9 | **职业骗保** | 特定高风险职业频繁出险(矿工/高危行业) |\r
| 10 | **重大事故造假** | 死亡/伤残事故真实性存疑 |\r
\r
#### 欺诈风险量化评分模型\r
\r
```python\r
# Fraud Risk Scoring Model (simplified / 简化示例)\r
def fraud_score(case):\r
    score = 0\r
    \r
    # 等待期观察(权重30%)\r
    if case.waiting_period_days \x3C 30:\r
        score += 30\r
    elif case.waiting_period_days \x3C 90:\r
        score += 15\r
    \r
    # 投保保额异常(权重25%)\r
    if case.coverage_to_income_ratio > 10:\r
        score += 25\r
    elif case.coverage_to_income_ratio > 5:\r
        score += 15\r
    \r
    # 就诊医院异常(权重20%)\r
    if case.hospital_fraud_history:\r
        score += 20\r
    elif case.is_chain_hospital == False and case.amount > 50000:\r
        score += 10\r
    \r
    # 既往理赔记录(权重15%)\r
    if case.past_fraud_record:\r
        score += 15\r
    elif case.past_claim_count > 5:\r
        score += 8\r
    \r
    # 同案关联(权重10%)\r
    if case.linked_to_suspected_cases:\r
        score += 10\r
    \r
    # 返回风险等级\r
    if score >= 60: return "HIGH_RISK - 立即调查"\r
    elif score >= 30: return "MEDIUM_RISK - 重点审核"\r
    else: return "LOW_RISK - 正常流程"\r
```\r
\r
### 3. Underwriting Risk Control / 核保风控体系\r
\r
#### 核保风险分级表\r
\r
| 风险等级 | 评估条件 | 核保决策 | 附加条件 |\r
|---------|---------|---------|---------|\r
| **标准体** | 健康告知全否、无异常记录 | 标准承保 | - |\r
| **次标准体A** | 轻微异常(BMI偏高/脂肪肝) | 加费承保 | +10%-30%保费 |\r
| **次标准体B** | 中等异常(高血压/糖尿病早期) | 有条件承保 | 除外责任+加费 |\r
| **延期体** | 病情不稳定或待确诊 | 延期观察 | 6-12个月后重核 |\r
| **拒保体** | 重大既往症/高风险职业 | 拒保 | 不予承保 |\r
\r
### 4. Fraud Investigation Workflow / 欺诈案件调查流程\r
\r
```text\r
受理报案\r
    ↓\r
初步审核(24小时内)\r
  ├─ 资料完整性检查\r
  ├─ 基本逻辑校验\r
  └─ 欺诈特征快速筛查(AI模型打分)\r
         ↓\r
风险分级\r
  ├─ LOW_RISK → 正常理赔流程\r
  ├─ MEDIUM_RISK → 理赔调查(7天内)\r
  └─ HIGH_RISK → 深度调查+报案\r
         ↓\r
调查阶段\r
  ├─ 医疗数据核实(医保/医院调取)\r
  ├─ 既往投保记录查询(银保信平台)\r
  ├─ 现场走访/面访\r
  └─ 第三方数据核查(大数据风控)\r
         ↓\r
结论处置\r
  ├─ 正常赔付 → 快速支付\r
  ├─ 协议赔付 → 减少金额达成和解\r
  ├─ 拒赔处理 → 发拒赔通知+说明\r
  └─ 报案追究 → 移送公安(涉嫌犯罪)\r
```\r
\r
### 5. Anti-Fraud Capability Benchmark / 反欺诈能力建设\r
\r
| 指标 | 行业平均水平 | 先进水平 | 说明 |\r
|------|------------|---------|------|\r
| **欺诈渗漏率** | 15%-20% | \x3C5% | 欺诈+渗漏(过度理赔)占保费比例 |\r
| **欺诈识别率** | 30%-40% | >70% | 实际欺诈案件中被识别的比例 |\r
| **案件调查周期** | 15-30天 | \x3C7天 | 从报案到结论 |\r
\r
---\r
\r
## Reference Files / 参考文件\r
\r
| File / 文件 | Content / 内容说明 |\r
|------|---------|\r
| `references/anti_fraud_guide.md` | 反欺诈工作办法解读 + 各险种欺诈特征详解 |\r
| `references/underwriting_risk_assessment.md` | 核保风控体系,含风险分级表和异常标记规则 |\r
| `references/claim_investigation_sop.md` | 理赔欺诈调查 SOP,含调查话术和流程图 |\r
| `references/fraud_case_study.md` | 典型欺诈案例分析(含真实案例改编) |\r
| `references/data_model_guide.md` | 反欺诈数据模型构建指南,含评分卡示例 |\r
Usage Guidance
This skill appears safe to install as an instruction-only reference skill. Before using it on real cases, confirm the legal basis for any medical or insurance data used, redact unnecessary personal information, verify the missing reference materials if needed, and keep human claims, underwriting, and compliance reviewers responsible for final decisions.
Capability Analysis
Type: OpenClaw Skill Name: insurance-anti-fraud Version: 1.1.0 The 'Insurance Anti-Fraud Expert' skill is a domain-specific tool designed to assist in identifying insurance fraud based on Chinese regulatory frameworks (CBIRC 2024). The content across SKILL.md, README.md, and the reference files consists of educational material, risk scoring logic, and procedural workflows for insurance claims and underwriting. There is no evidence of malicious intent, data exfiltration, or unauthorized command execution; the provided Python snippet is a simplified logic example for risk scoring and does not pose a security threat.
Capability Assessment
Purpose & Capability
The stated purpose and content are coherent: the skill provides insurance anti-fraud, claims, underwriting, and investigation guidance. Because it can influence claim review, refusal, underwriting, or police-referral decisions, outputs should be reviewed by qualified compliance and claims staff.
Instruction Scope
The workflow discusses checking medical, insurance, and third-party data sources. That is purpose-aligned for fraud investigation, but users should ensure legal authorization and data minimization before providing case data to an agent.
Install Mechanism
There is no install spec and no code files. The README contains a normal user-directed install command, but several reference files advertised in SKILL.md are not present in the supplied manifest.
Credentials
The artifacts declare no required binaries, environment variables, credentials, config paths, network integrations, or local system access.
Persistence & Privilege
No persistence, background agent behavior, memory storage, privilege escalation, or autonomous execution mechanism is shown.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install insurance-anti-fraud
  3. After installation, invoke the skill by name or use /insurance-anti-fraud
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.1.0
v1.1.0 Bilingual optimization: English metadata + English summaries; Bilingual README.md; SEO title for anti-fraud analysis and CBIRC compliance.
v1.0.1
v1.0.1: 完善欺诈评分卡,增加大数据反欺诈技术路线说明
v1.0.0
ClawHub首个中国保险反欺诈垂直Skill,覆盖2024年银保监会《反保险欺诈工作办法》、理赔欺诈识别引擎、核保风控体系、欺诈评分模型。行业欺诈渗漏率15-20%,先进水平5%以下
Metadata
Slug insurance-anti-fraud
Version 1.1.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 3
Frequently Asked Questions

What is Insurance Anti-Fraud Expert?

基于国家监管办法,提供保险理赔和核保欺诈识别、风险评分、调查流程及防范建议,助力风控合规。 It is an AI Agent Skill for Claude Code / OpenClaw, with 102 downloads so far.

How do I install Insurance Anti-Fraud Expert?

Run "/install insurance-anti-fraud" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Insurance Anti-Fraud Expert free?

Yes, Insurance Anti-Fraud Expert is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Insurance Anti-Fraud Expert support?

Insurance Anti-Fraud Expert is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Insurance Anti-Fraud Expert?

It is built and maintained by gechengling (@gechengling); the current version is v1.1.0.

💬 Comments