Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

医学量表检索技能

v1.0.0

医学量表信息检索与标准化报告生成。通过浏览器自动化检索量表背景、版权、CDISC 映射、统计方法等信息,生成飞书云文档并保存到量表知识库。

0· 110·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for kevinget-svg/medical-scale-research.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "医学量表检索技能" (kevinget-svg/medical-scale-research) from ClawHub.
Skill page: https://clawhub.ai/kevinget-svg/medical-scale-research
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Canonical install target

openclaw skills install kevinget-svg/medical-scale-research

ClawHub CLI

Package manager switcher

npx clawhub@latest install medical-scale-research
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill's purpose (retrieving scale info, getting PDFs/screenshots, generating Feishu docs, CDISC mapping) matches the instructions: it uses browser automation, web fetching and Feishu APIs. However, the metadata declares no required environment variables or credentials even though the README and SKILL.md explicitly require Feishu API permissions and browser automation privileges. Also the SKILL.md hard-codes a local path (/Users/wangyafei/Downloads/...) which is disproportionate and suggests the package was authored against a specific developer environment.
!
Instruction Scope
Runtime instructions direct the agent to download official PDFs using an exec/curl command, verify local files (ls -lh), take screenshots (browser.snapshot) and upload them to Feishu. These steps involve filesystem writes, shell execution, and automated browsing; they go beyond read-only retrieval and require explicit user consent and credentials. The instructions also force 'must download PDF or screenshot' as a precondition, which could cause the agent to attempt network downloads or local writes even when access is restricted or when content is paywalled.
Install Mechanism
This is instruction-only (no install spec, no code files to execute), which minimizes supply-chain risk. No external archives or unusual installers are pulled in. The attack surface is primarily from the agent executing the described runtime actions, not from a downloaded binary.
!
Credentials
The skill requires Feishu API access and browser automation to work (per README and SKILL.md) but declares no required environment variables or primary credential in the registry metadata; that mismatch is problematic. It also mandates writing files into a hard-coded user path and uses shell commands (curl, ls) without declaring required binaries. These requirements imply access to local filesystem, network, and third-party service credentials that are not surfaced in the metadata.
Persistence & Privilege
always:false and no modifications to other skills or global agent config are requested. The skill does not claim permanent inclusion or elevated platform-wide privileges.
What to consider before installing
What to check before installing/using this skill: - Credentials and permissions: The skill expects Feishu API access and browser automation privileges, but the registry entry lists no required environment variables or primary credential. Ask the publisher how Feishu tokens/credentials are provided and where they are stored; do not hand over global credentials without understanding scope and storage. - Local filesystem writes and shell execution: The runtime instructions instruct the agent to run shell commands (curl, ls) and save PDFs to a hard-coded path (/Users/wangyafei/Downloads/...). If you install this skill, change default paths to a safe, configurable folder and confirm the agent is allowed to write there. Prefer running in an isolated environment or sandbox during initial testing. - Forced downloads: The skill enforces mandatory retrieval of the 'official PDF' (download or screenshot). Ensure you have legal right to download/archive the PDF (copyright/commercial restrictions); otherwise the agent may attempt to access paywalled or restricted content. - Limit autonomous actions: The skill can be invoked by the agent autonomously (default). If you are uncomfortable with automated browsing + shell downloads, require user consent for network/file operations or disable autonomous invocation until you verify behavior. - Verify implementation details: Request clarification from the maintainer about how Feishu credentials are supplied, whether tokens are transient, and whether the skill will ever send data to external endpoints other than Feishu and the public sources named. Ask them to remove hard-coded user paths and to declare required env vars (Feishu token, any proxy settings) in the metadata. - Test in a sandbox first: Run the skill in a restricted test account and sandboxed filesystem, with minimized credentials, to observe exactly what it does before granting it access to production data/accounts. If the publisher cannot or will not clarify these points (where credentials come from, configurable save path, and why no env vars are declared), treat the skill as high-risk and avoid using it with sensitive accounts or systems.

Like a lobster shell, security has layers — review code before you run it.

CDISCvk977z2q6rh0pqy9ff0f7tctxb1854343clinical-trialsvk977z2q6rh0pqy9ff0f7tctxb1854343feishuvk977z2q6rh0pqy9ff0f7tctxb1854343latestvk977z2q6rh0pqy9ff0f7tctxb1854343medicalvk977z2q6rh0pqy9ff0f7tctxb1854343openclawvk977z2q6rh0pqy9ff0f7tctxb1854343scalevk977z2q6rh0pqy9ff0f7tctxb1854343skillvk977z2q6rh0pqy9ff0f7tctxb1854343
110downloads
0stars
1versions
Updated 1w ago
v1.0.0
MIT-0

医学量表 Skill - medical-scale-research

PURPOSE

通过浏览器自动化检索医学量表的完整信息(背景、版权、CDISC 映射、统计方法等),生成标准化的飞书云文档,并保存到量表知识库。支持版本追踪和增量更新。

WHEN TO USE

TRIGGERS

  • "检索 [量表名] 量表的信息"
  • "整理 [量表名] 量表的报告"
  • "查询 [量表名] 的 CDISC domain"
  • "更新 [量表名] 量表的版本信息"
  • "[量表名] 量表的统计分析方法是什么"

DO NOT USE WHEN

  • 需要访问付费墙后的完整文献(只能获取摘要)
  • 量表信息完全不存在于公开网络
  • 用户需要即时版权授权(需联系版权方)

INPUTS

REQUIRED

  • 量表名称(中文或英文,如 "SGRQ" / "圣乔治呼吸问卷")

OPTIONAL

  • 指定版本(如 "最新版本" 或 "v2.0")
  • 指定检索重点(如 "重点检索 CDISC 映射")
  • 更新模式(如 "增量更新已有文档")

EXAMPLES

  • "检索 SGRQ 量表的信息"
  • "整理 EORTC QLQ-C30 量表的完整报告"
  • "查询 PHQ-9 量表的 CDISC domain 和统计方法"
  • "更新 EASI 量表的版本历史"

OUTPUTS

默认输出

  • 飞书云文档链接(标准化报告)
  • 知识库归档确认

文档结构

# [量表名称] 标准化研究报告

## 一、量表简介
### 1.1 背景及发展历程
### 1.2 应用领域
### 1.3 量表内容
### 1.4 版本历史(最新 + 更新摘要)
### 1.5 官方量表全文 ⚠️(必须有 PDF 链接或截图)

## 二、版权与授权信息
### 2.1 版权方信息
### 2.2 中文版信息

## 三、CDISC 编程
### 3.1 标准 aCRF 与受控术语(自动推荐 Domain)
### 3.2 SDTM 实现
### 3.3 ADaM 实现

## 四、统计分析
### 4.1 数据类型与分布特征
### 4.2 缺失值处理(条目级 + 访视级)
### 4.3 常规分析模型(含 SAS/R 程序)
### 4.4 探索性分析
### 4.5 结果呈现(Table/Figure 模板)

## 五、参考文献

1.5 官方量表全文章节模板

### 1.5 官方量表全文

**获取状态**:✅ 已下载 PDF / ✅ 已截图

**PDF 下载**:
- 下载链接:[量表名称 Official PDF](URL)
- 本地路径:`/Users/wangyafei/Downloads/scales/[量表名]_official.pdf`
- 文件大小:XX KB

**量表条目预览**:
(如为截图,此处插入图片)

| 部分 | 条目示例 | 评分方式 |
|------|---------|---------|
| ... | ... | ... |

> **注意**:完整量表请下载上方 PDF 文件。量表版权归 [版权方] 所有,仅限学术研究使用。

WORKFLOW

1. 解析量表名称

  • 提取关键词(中英文全称、缩写)
  • 识别同义词(如 "SGRQ" = "圣乔治呼吸问卷" = "St George's Respiratory Questionnaire")

2. 并行检索 7 个维度 + 量表全文获取

使用 browser 工具打开多个 Tab 并行检索:

维度数据源检索策略
背景信息PubMed, Google Scholar, 量表官网"[量表名] development validation history"
官方量表全文官网 PDF 下载链接"[量表名] official instrument PDF download"
版本历史官网,GitHub, 修订记录"[量表名] version history revision changelog"
版权信息Mapi Research Trust, 版权方官网"[量表名] copyright license permission"
中文版信息CNKI, 万方,中文论文"[量表名] 中文版 信效度 验证"
CDISC 映射CDISC 官网,TAUGs, SDTMIG"[量表名/概念] CDISC SDTM ADaM domain"
统计方法高引文献,系统评价"[量表名] scoring algorithm missing data analysis"

2.1 ⚠️ 量表全文获取(强制要求)

必须完成以下任一操作,否则停止执行并告知用户

选项 A:下载官方 PDF(优先)

1. 在官网/权威来源查找 PDF 下载链接
2. 使用 exec 下载:curl -L -o "/Users/wangyafei/Downloads/scales/[量表名]_official.pdf" "[PDF_URL]"
3. 验证文件:ls -lh 确认文件存在且>1KB
4. 在文档 1.4 章节后插入"官方量表全文"章节,包含:
   - PDF 下载链接(可点击)
   - 本地保存路径
   - 文件大小
   - 版权说明

选项 B:屏幕截图(无法下载时)

1. 打开量表官方页面(browser.open)
2. 使用 browser.snapshot 捕获页面(aria refs)
3. 如页面较长 → 滚动分多次截图(至少覆盖完整量表条目)
4. 使用 feishu_doc.upload_image 上传图片到飞书云盘
5. 在文档 1.4 章节后插入"官方量表全文"章节,包含:
   - 量表条目截图(至少 1 张)
   - 来源网址链接
   - 版权说明

❌ 无法获取时的处理

如无法找到 PDF 且无法截图:
1. 停止执行
2. 告知用户:"未找到 [量表名] 的官方量表全文(PDF 或截图)"
3. 提供版权方联系方式
4. 询问用户是否继续生成报告(不含量表全文)

2.2 量表全文检索策略

检索关键词(按顺序尝试):

1. "[量表名] official PDF download"
2. "[量表名] full instrument PDF"
3. "[量表名] questionnaire PDF"
4. "[量表名] 官网" → 查找下载链接
5. "[量表名] Mapi Research Trust" → 版权方官网

权威来源优先级

1. 量表官方网站(.edu/.org 域名)
2. 版权方官网(如 Mapi Research Trust)
3. 原作者机构网站
4. 指南/共识文件(如 EPOS2020)
5. 学术论文补充材料(Supplementary Materials)

3. 信息提取与结构化

  • 使用 web_fetch 提取页面内容
  • 过滤噪声,提取关键字段
  • 验证信息一致性(多源交叉验证)

4. CDISC 自动推荐

基于规则引擎推荐 Domain:

{
  "生活质量/HRQoL": {"sdtm": "QS", "adam": "ADQS"},
  "症状评估": {"sdtm": "QS", "adam": "ADQS"},
  "病史": {"sdtm": "MH", "adam": "ADMH"},
  "合并用药": {"sdtm": "CM", "adam": "ADCM"},
  "不良事件": {"sdtm": "AE", "adam": "ADAE"},
  "生命体征": {"sdtm": "VS", "adam": "ADVS"},
  "实验室检查": {"sdtm": "LB", "adam": "ADLB"}
}

5. 生成飞书文档(重要!)

⚠️ 内容长度限制:飞书 API 对单次写入内容长度有限制(约 5KB)

推荐方法

IF 内容 > 5KB:
  1. 先 create 创建空文档
  2. 将内容按章节分割(每段<4KB)
  3. 使用 insert 操作逐段插入(after_block_id)
  4. 每插入一段后获取新的 block_id
ELSE:
  直接使用 write 操作一次性写入

操作步骤

  1. 创建文档feishu_doc create → 获取 doc_token
  2. 写入第一章feishu_doc write → 写入完整第一章
  3. 获取最后 block_idfeishu_doc list_blocks → 获取最后一个 block_id
  4. 插入后续章节feishu_doc insert + after_block_id → 逐章插入
  5. 验证内容feishu_doc read → 确认内容完整

内容分割策略

章节 1:一、量表简介(约 3KB)
章节 2:二、版权与授权信息(约 2KB)
章节 3:三、CDISC 编程(约 3KB)
章节 4:四、统计分析(约 4KB)
章节 5:五、参考文献 + 附录(约 2KB)

6. 保存到知识库

  • 使用 feishu_wiki 创建/更新节点
  • 按疾病领域/量表类型分类
  • 记录版本历史

7. 增量更新(可选)

  • 对比新旧版本
  • 高亮变更内容
  • 更新版本历史

SAFETY & EDGE CASES

  • 版权信息:明确标注信息来源,不提供法律建议
  • 版本准确性:优先使用最新版本,标注检索日期
  • CDISC 推荐:标注为"推荐",需人工确认
  • 付费内容:只能获取摘要,标注"需订阅"
  • 多版本并存:明确标注各版本差异
  • 信息冲突:多源交叉验证,标注不确定性
  • ⚠️ 飞书文档长度限制
    • 单次写入内容 < 5KB
    • 长文档需分章节插入
    • 使用 insert + after_block_id 逐段写入
  • ⚠️ 量表全文获取(强制)
    • 必须下载 PDF 或截图,否则停止执行
    • PDF 文件需验证存在且>1KB
    • 截图需覆盖完整量表条目

STOP AND ASK THE USER IF

  • 量表名称模糊/有多个同名量表
  • 版权信息无法确认(需用户核实)
  • CDISC 推荐置信度低(需人工判断)
  • 检索结果为空(需调整策略)
  • 知识库权限不足(需用户授权)
  • ⚠️ 无法获取量表全文(PDF/截图)
    • 停止执行并告知用户
    • 提供已尝试的检索来源
    • 提供版权方联系方式
    • 询问是否继续生成报告(不含量表全文)

EXAMPLES

Example 1: 首次检索

Input: "检索 SGRQ 量表的信息"

Process:

  1. 解析 → SGRQ = St George's Respiratory Questionnaire = 圣乔治呼吸问卷
  2. 并行检索 7 个维度
  3. 提取结构化数据
  4. CDISC 推荐 → QS (生活质量)
  5. 生成飞书文档
  6. 保存到知识库

Output:

✅ 已完成 SGRQ 量表信息检索

📄 文档链接:https://xxx.feishu.cn/docx/xxxxx
📚 已归档至:医学量表库 > 呼吸系统 > SGRQ

检索维度:
✓ 背景信息(5 篇文献)
✓ 官方量表(✅ 已下载 PDF - 7.0KB)
✓ 版本历史(v2.1 最新版,2023 更新)
✓ 版权信息(Mapi Research Trust)
✓ 中文版(3 篇验证文献)
✓ CDISC 映射(QS/ADQS 推荐)
✓ 统计方法(4 种模型)

📎 量表全文:
- PDF 路径:/Users/wangyafei/Downloads/scales/SGRQ_official.pdf
- 下载链接:[SGRQ Official PDF](https://xxx.xxx.ac.uk/xxx.pdf)

Example 2: 增量更新

Input: "更新 EASI 量表的版本信息"

Process:

  1. 读取已有文档
  2. 检索最新版本信息
  3. 对比变更内容
  4. 更新版本历史章节
  5. 标注变更日期

Output:

✅ 已更新 EASI 量表版本信息

📝 新增内容:
- v2024 版本(2024-01 发布)
- 儿童权重系数调整
- 新增深色皮肤红斑评估指南

📄 更新链接:https://xxx.feishu.cn/docx/xxxxx

FILES

  • templates/scale-report.md - 文档模板
  • rules/cdisc-mapping.json - CDISC 推荐规则
  • examples/SGRQ-example.md - 示例输出

RELATED SKILLS

  • lark-doc - 飞书文档操作
  • lark-wiki - 飞书知识库管理
  • agent-browser - 浏览器自动化

Comments

Loading comments...