← 返回 Skills 市场

保险条款分析

Name: 保险条款分析
Author: long1973m

作者马龙 · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ✓ 安全检测通过

289

总下载

当前安装

版本数

在 OpenClaw 中安装

/install insurance-policy-parser

功能描述

解析医疗险条款文档并提取32个结构化字段，含14个核心字段和18个增强字段，支持PDF/DOCX/TXT格式，输出标准JSON

使用说明 (SKILL.md)

医疗险条款结构化解析器

任务目标

本 Skill 用于：从医疗险条款文档中提取32个结构化字段
能力包含：解析PDF/DOCX/TXT格式文档，提取Level 1核心字段（14个）和Level 2增强字段（18个）
触发条件：用户提供医疗险条款文档或文本内容，需要结构化提取时

前置准备

依赖说明：无特殊依赖，已内置文档解析脚本

操作步骤

标准流程:
1. 获取文档内容
  - 如果用户提供的是PDF/DOCX/TXT文件，调用 scripts/parse_document.py \x3Cfile_path> 获取纯文本内容
  - 如果用户直接输入保险条款文本，直接使用该文本
2. 提取结构化信息
  - 根据 references/output-format.md 中的字段定义，从文档内容中提取对应信息
  - 智能体将逐项分析文档，填充以下两部分：
    - level1_core_fields: 14个绝对核心字段（年度限额、免赔额、赔付比例、续保条件等）
    - level2_enhancement_fields: 18个对比增强字段（家庭共享免赔额、特殊门诊、院外购药、CAR-T等）
  - 数据类型转换规则：
    - DECIMAL类型：金额转换为元（如"100万" → 1000000.00），比例转换为小数（如"100%" → 1.0000）
    - ENUM类型：严格使用枚举值（如免赔额单位："年"/"次"/"疾病"/"住院"）
    - BOOLEAN类型：根据表述判断（"保障"/"是" → true，"不保障"/"否" → false）
    - INT类型：直接提取数字（如"30天" → 30）
3. 输出结果
  - 以JSON格式输出完整的结构化数据
  - Level 1字段必须填充，Level 2字段未明确说明可使用null
  - 确保所有字段类型和枚举值符合格式规范

资源索引

必要脚本:见 scripts/parse_document.py(用途与参数:解析PDF/DOCX/TXT文件，返回纯文本)
领域参考:见 references/output-format.md(何时读取:始终读取，用于了解字段定义和提取规则)

注意事项

优先使用文档解析脚本处理文件，避免格式错误
提取时保持原文含义，必要时可进行总结归纳
Level 1核心字段必须优先保证准确性，AI可100%提取无null风险
Level 2增强字段允许null，体现产品差异化
严格按照数据类型转换规则处理数值、比例、枚举值
确保JSON格式正确，避免语法错误

使用示例

场景1：上传PDF医疗险合同
- 调用脚本：python scripts/parse_document.py ./insurance_contract.pdf
- 提取32个字段并输出结构化JSON
- 核心字段示例：年度限额200万元、免赔额1万元/年、社保报销后100%赔付
- 增强字段示例：院外购药保障、CAR-T治疗、家庭共享免赔额
场景2：直接输入医疗险条款文本
- 直接分析文本内容
- 按格式规范输出32个字段的结构化JSON
- 未明确说明的Level 2字段使用null表示

安全使用建议

This skill appears coherent and focused on extracting structured fields from user-supplied insurance documents. Before installing/using it: (1) recognize it will read the full text of any uploaded document — do not supply documents containing personal/medical data unless you trust where the agent runs and how outputs are stored/transmitted; (2) ensure your environment has the required Python libraries (PyPDF2 and python-docx) or expect ImportError; (3) note it does not perform OCR on scanned images—provide selectable text PDFs or run OCR beforehand; (4) test with benign documents to confirm output matches the JSON schema and to validate edge cases; (5) be skeptical of the claimed '100% extraction' accuracy — manually review outputs for correctness.

功能分析

Type: OpenClaw Skill Name: insurance-policy-parser Version: 1.0.0 The skill bundle is a legitimate tool for parsing insurance policy documents (PDF, DOCX, TXT) and extracting structured data into JSON. The Python script `scripts/parse_document.py` uses standard libraries (PyPDF2, python-docx) to extract text and contains no network calls, obfuscation, or unauthorized file access. The instructions in `SKILL.md` and `references/output-format.md` are strictly focused on the data extraction task and do not contain any prompt injection attempts or malicious commands.

能力评估

✓ Purpose & Capability

Name/description describe extracting 32 fields from medical-insurance documents; the repository contains a small Python script that extracts text from PDF/DOCX/TXT and a detailed output-format spec. Declared dependencies (PyPDF2, python-docx) match the parser's needs. No unrelated credentials, binaries, or install steps are requested.

ℹ Instruction Scope

SKILL.md limits runtime behavior to: (1) run scripts/parse_document.py on user-supplied files or parse provided text, (2) map extracted text to the JSON schema in references/output-format.md. This stays within scope. Note: the README asserts the AI can '100% extract' Level 1 fields — that is an overconfident claim (extraction quality depends on document layout, OCR needs for scanned PDFs, ambiguous language, etc.). The skill does not handle OCR or images; it only extracts text content.

✓ Install Mechanism

No install spec provided (instruction-only), so nothing is downloaded or executed beyond the included parse_document.py. Dependencies are declared in SKILL.md but not automatically installed by the skill; missing libraries will cause the script to raise ImportError. No external URLs, archives, or unusual install locations are used.

✓ Credentials

The skill requires no environment variables, credentials, or config paths. The requested resources are proportional to its function (text parsing of user-provided documents).

✓ Persistence & Privilege

The skill does not request always:true or other elevated persistence. It contains no code that modifies other skills or system-wide configs; it simply reads input files and prints extracted text.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install insurance-policy-parser
安装完成后，直接呼叫该 Skill 的名称或使用 /insurance-policy-parser 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release: Extracts 32 structured fields from medical insurance policy documents in PDF, DOCX, or TXT formats, outputting standardized JSON. - Supports both core (14) and enhancement (18) fields as defined in the output format specification. - Implements automated data type conversion for amounts, proportions, enumerations, and booleans. - Provides a document parsing script for robust text extraction from files. - Ensures all Level 1 core fields are mandatory and accurately extracted; Level 2 enhancement fields allow nulls. - Output strictly follows predefined JSON formatting and field rules.

元数据

Slug insurance-policy-parser

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题