Body Report Interpreter

解读体检报告（血常规、肝功能、血脂、肾功能、尿常规、肿瘤标志物等）。用户上传体检报告图片或PDF时触发。逐项解释指标含义、标注异常值、给出健康建议。适用场景：解读体检报告、解释化验单、健康指标异常分析、历年体检对比。

Security Scan

The skill's instructions, files, and requirements are coherent with its stated purpose of interpreting medical test reports and do not request unexpected credentials, installs, or system access.

Purpose & Capability

The name/description (interpreting blood tests, liver, lipids, kidney, urine, tumor markers) matches the provided SKILL.md and references/ranges.md. The skill requires no binaries, environment variables, or installs — consistent with an instruction-only parser/formatter that uses OCR/read on user-uploaded reports.

Instruction Scope

Instructions explicitly instruct the agent to read user-uploaded images/PDFs (via a 'read' tool) and parse personal report fields (date, hospital, name). This is expected for the feature but means the skill will handle sensitive personal health information (PHI). The SKILL.md does not instruct sending data to external endpoints, which is good, but the runtime behavior of the unspecified 'read' tool (local vs. external OCR) is not described and could affect privacy. Also watch for OCR/unit mismatches and reliance on default ranges — SKILL.md already notes preferring the report's own ranges first.

Install Mechanism

No install specification and no code files — instruction-only. This minimizes risk (nothing is written to disk or fetched at install).

Credentials

The skill declares no required environment variables, credentials, or config paths. For the stated functionality (parsing and explaining lab values) this is appropriate and not excessive.

Persistence & Privilege

Flags show always: false and normal invocability. The skill does not request persistent system privileges or to modify other skills/config. This level of presence is proportionate.

Assessment

This skill appears coherent and limited to parsing and explaining lab reports, but it will process sensitive health data. Before installing or using it: (1) confirm where OCR/‘read’ processing runs (locally vs. external service); prefer local processing if possible to protect privacy; (2) avoid uploading other people's reports without consent; (3) verify that the skill uses the report's own reference ranges (it states it will) and check units/sex-specific ranges manually when in doubt; (4) treat outputs as informational, not a medical diagnosis — follow up with a clinician for serious or red-flag results; (5) monitor for future updates that add network endpoints, environment variables, or install steps — those would materially change the risk profile. If you need a higher-assurance evaluation, provide runtime details for the 'read' tool and any telemetry/network behavior the agent may perform.

体检报告解读

核心流程

1. 获取报告内容

• 用户上传图片/PDF → 使用 read 工具读取
• 如用户粘贴文字 → 直接解析
• 识别报告日期、医院、姓名等基本信息

2. 逐项解读

对每个指标执行：

1. 指标名称 → 中文+英文缩写
2. 检测值 → 用户的实际数值
3. 参考范围 → 正常区间（见 references/ranges.md）
4. 状态 → ⬆️偏高 / ✅正常 / ⬇️偏低
5. 含义 → 一句话说明这个指标反映什么

3. 异常指标重点分析

对所有异常指标：

• 可能原因（列出2-3个常见原因）
• 是否需要就医（分等级：🔴 建议尽快 / 🟡 建议复查 / 🟢 注意观察）
• 饮食/生活建议

4. 总结

• 整体健康评估（一句话）
• 需要关注的重点项
• 建议复查项目和时间

输出格式

📋 体检报告解读

📅 报告日期：YYYY-MM-DD
🏥 医院：XXX

━━━ 项目分类 ━━━

| 指标 | 结果 | 参考范围 | 状态 |
|------|------|----------|------|
| XXX  | XXX  | XXX-XXX  | ✅/⬆️/⬇️ |

⚠️ 异常指标详解

1. 【指标名】 ⬆️
   - 值：XXX（参考：XXX-XXX）
   - 可能原因：...
   - 建议：...

💡 整体建议
...

重要规则

• 不做诊断：只解读指标含义，不说"你得了XX病"
• 建议就医：明显异常时建议看医生，附带建议科室
• 参考范围：优先使用报告上自带的范围，其次用 references/ranges.md
• 通俗易懂：避免医学术语堆砌，用大白话解释