Skill flagged — suspicious patterns detected

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

血糖管理助手

v1.0.0

管理个人血糖数据,支持手动记录、文件导入、血糖趋势预测和健康建议。当用户提到血糖、糖尿病管理、血糖记录、血糖预测、饮食建议、运动建议时使用此技能。

0· 97·0 current·0 all-time
bylining@liningg

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for liningg/glucose-manager.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "血糖管理助手" (liningg/glucose-manager) from ClawHub.
Skill page: https://clawhub.ai/liningg/glucose-manager
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

Bare skill slug

openclaw skills install glucose-manager

ClawHub CLI

Package manager switcher

npx clawhub@latest install glucose-manager
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
The skill claims to store all data locally and only provide analysis/predictions, which matches the included scripts, dependencies, and no external install URLs. However, there are mismatches between components: SKILL.md/README and data_manager refer to '~/.workbuddy/data/glucose_readings.json', while advice_generator.py uses '~/.workbuddy/glucose_data.json'. Several scripts expect different JSON field names (e.g., data_manager and many analysis scripts use 'value' and a 'readings' array, whereas advice_generator refers to 'glucose_value' and 'records'/'user_profile'). These inconsistent defaults and schemas mean different parts of the package may not interoperate and could create unexpected files or lost data. This is disproportionate to the stated purpose and should be fixed before use.
!
Instruction Scope
SKILL.md instructs the agent to run local scripts (record, import, analyze, predict, recommend), which aligns with available scripts. But instructions assert that 'All data stored locally' and list a single canonical storage path that is not consistently honored by the code. The runtime examples and scripts reference local filesystem paths only (no explicit external endpoints in shown files), which is appropriate; however the mismatch in filenames/keys grants the agent broad discretion to read/write multiple locations and may produce or read unexpected files. Also the SKILL.md contains detected unicode control characters (prompt-injection signal) which could be used to manipulate runtime parsing—this is unexpected for a medical-recording skill.
Install Mechanism
No install spec (instruction-only at packaging level), so nothing will be implicitly downloaded at install time. The included skill.json declares typical Python data-analysis dependencies (pandas, openpyxl, matplotlib/plotly, numpy) which are appropriate for this functionality. There are no remote download URLs or unusual installers observed in the provided excerpts.
Credentials
The skill does not request any environment variables, credentials, or config paths beyond local user-home storage. That is proportionate for a local glucose manager. No evidence in the visible files of requests for unrelated secrets.
Persistence & Privilege
always is false and the skill does not request elevated privileges. The package writes to user-local directories (~/.workbuddy), creates backups, and manages files within that directory — reasonable for its purpose. It does not declare any behavior that modifies other skills or system-wide agent settings in the provided excerpts.
Scan Findings in Context
[unicode-control-chars] unexpected: SKILL.md contained unicode control characters detection. This is not expected for a straightforward local health assistant and may indicate an attempt to influence prompt parsing or evaluation. It is not evidence of data exfiltration by itself, but warrants caution and manual inspection of the SKILL.md and any parsing logic that consumes it.
What to consider before installing
What to consider before installing: 1) Inconsistent local storage and schemas: Several files disagree about the canonical storage path and JSON schema (examples: data_manager stores ~/.workbuddy/data/glucose_readings.json with 'readings' and 'value', while advice_generator defaults to ~/.workbuddy/glucose_data.json and looks for 'glucose_value' and 'records'). This will likely cause parts of the skill to read an empty file or create separate files. Ask the author or inspect/standardize the code (pick one path/format) before trusting your real health data. 2) Audit the omitted files for network activity: The provided excerpts show no network calls, but 21 files were truncated in the package. Before use, grep the full code for networking libraries or calls (requests, urllib, socket, http.client, ftplib, boto3, paramiko, websockets, aiohttp, smtplib) and for hardcoded URLs or IPs. If any are present, get an explanation and justification. 3) Fix schema mismatches or run in a sandbox: Because scripts use different field names and locations, run the skill first in an isolated environment (or a disposable account) and verify where files are written and what keys are used. Backup any existing ~/.workbuddy data before running. 4) Prompt-injection artifact: The SKILL.md included unicode-control characters flagged by static scanning. That could be benign (formatting artifacts) but may also be used to manipulate parsers. Manually inspect the SKILL.md for hidden characters and remove them or ask the author to provide a clean copy. 5) Privacy claims vs. reality: The README/SKILL.md claim 'no external transmission'. Confirm by static audit (search for network I/O) and by monitoring network activity when running the skill (e.g., in a sandbox) if you plan to store real personal health data. 6) Reasonable dependencies but install carefully: The dependency list is appropriate for offline analysis, but these Python packages should be installed from trusted package repositories. Prefer creating a virtual environment. 7) If you lack time/skill to audit: Consider not installing or only use after the author addresses the data-path and schema inconsistencies and provides assurance (or a signed release) that no telemetry exists. If you proceed, keep sensitive or real patient data out until you've validated behavior.

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

latestvk97cx4w2bskmzme1ck76nnndk583hcag
97downloads
0stars
1versions
Updated 1mo ago
v1.0.0
MIT-0

血糖管理助手

概述

这是一个综合性的血糖管理技能,帮助用户记录、分析和预测血糖数据,并提供个性化的健康建议。支持手动输入、文件导入等多种数据记录方式,提供血糖趋势预测、风险评估以及饮食、运动、用药和就医建议。

核心功能

1. 血糖数据管理

手动录入

  • 记录血糖值、时间戳、餐前/餐后状态和备注
  • 支持不同血糖单位(mmol/L 和 mg/dL)
  • 自动单位转换和标准化

文件导入

  • 从 CSV 文件导入血糖数据
  • 从 Excel 电子表格导入血糖数据
  • 验证数据格式并处理缺失值

数据存储

  • 所有血糖数据以 JSON 格式存储于 ~/.workbuddy/data/glucose_readings.json
  • 每次修改前自动备份
  • 数据结构包含:时间戳、血糖值、单位、餐前/餐后状态、备注、标签

2. 趋势分析与可视化

统计分析

  • 计算日、周、月血糖统计数据
  • 识别血糖模式(达标率、低血糖/高血糖频率)
  • 生成血糖波动指标(标准差、变异系数)

可视化报告

  • 使用 matplotlib/plotly 创建血糖趋势图表
  • 生成基于时间的血糖分布图
  • 导出 PNG 或 HTML 格式的报告

3. 预测建模

短期预测

  • 基于近期读数预测未来 2-4 小时血糖水平
  • 识别潜在低血糖/高血糖风险时段

模式识别

  • 检测周期性血糖模式(如餐后峰值、黎明现象)
  • 识别与血糖波动相关的因素

风险评估

  • 计算血糖异常风险评分
  • 基于趋势生成早期预警提醒

4. 个性化建议

饮食指导

  • 根据血糖模式建议碳水化合物管理策略
  • 推荐用餐时间调整
  • 提供血糖指数(GI)相关指导

运动建议

  • 根据血糖模式建议最佳运动时间
  • 推荐运动强度和持续时间
  • 警告运动诱发的低血糖风险

健康提醒

  • 当血糖模式提示需要就医时发出提醒
  • 建议何时需要更频繁监测血糖
  • 识别可能需要调整用药的模式

交互示例

以下是与用户的真实对话示例,展示了如何使用本技能:

📝 血糖记录场景

场景 1:简单记录

用户: 我血糖7.2
助手: ✅ 血糖记录已保存!
      🩸 血糖: 7.2 mmol/L
      📅 时间: 2026-03-24 16:30
      📊 状态: 略高于正常范围,建议关注

场景 2:空腹血糖

用户: 今天早餐前血糖6.9
助手: ✅ 已记录空腹血糖!
      🩸 血糖: 6.9 mmol/L
      🍽️ 状态: 空腹(早餐前)
      
      📊 空腹血糖参考:
      • 正常: 3.9-6.1 mmol/L
      • 糖前期: 6.1-7.0 mmol/L
      • 您处于糖前期临界,建议控制饮食

场景 3:餐后血糖

用户: 刚吃完午饭两小时,血糖9.5
助手: ✅ 餐后血糖已记录!
      🩸 血糖: 9.5 mmol/L
      🍽️ 状态: 餐后2小时
      
      📊 餐后血糖参考:
      • 正常: <7.8 mmol/L
      • 糖耐量受损: 7.8-11.1 mmol/L
      • 餐后偏高,建议减少精制碳水摄入

场景 4:带备注记录

用户: 刚运动完,血糖5.2,跑了30分钟
助手: ✅ 血糖记录已保存!
      🩸 血糖: 5.2 mmol/L
      📝 备注: 运动30分钟后
      ⚠️ 提示: 运动后血糖略低,注意补充能量

📊 数据查询场景

场景 5:查看报告

用户: 看看这周的血糖报告
助手: 📊 本周血糖报告(3/18-3/24)
      
      📈 数据概览:
      • 平均血糖: 7.2 mmol/L
      • 最高值: 11.2 mmol/L
      • 最低值: 5.4 mmol/L
      • 达标率: 72%
      • 监测次数: 27次
      
      💡 建议: 餐后血糖波动较大,建议控制碳水摄入

场景 6:趋势分析

用户: 最近血糖趋势怎么样
助手: 📈 血糖趋势分析
      
      🟢 好的方面:
      • 空腹血糖稳定,平均6.4 mmol/L
      • 无低血糖事件
      
      🟡 需关注:
      • 餐后血糖波动大,峰值达11.2
      • 3/22晚餐后血糖偏高
      
      💡 建议: 晚餐减少主食,增加蔬菜

场景 7:对比分析

用户: 这周和上周比怎么样
助手: 📊 周对比分析
      
      | 指标 | 上周 | 本周 | 变化 |
      |-----|------|------|-----|
      | 平均血糖 | 7.8 | 7.2 | ↓0.6 |
      | 达标率 | 65% | 72% | ↑7% |
      | 监测次数 | 21 | 27 | +6 |
      
      ✅ 本周血糖控制有进步,继续保持!

🔮 预测场景

场景 8:风险评估

用户: 今晚会不会低血糖
助手: 🔮 血糖风险评估
      
      📊 当前状态: 16:30 血糖 7.2 mmol/L
      🔮 今晚预测: 
      • 18:00 预计 6.5 mmol/L
      • 21:00 预计 5.8 mmol/L
      • 睡前 预计 5.5 mmol/L
      
      ⚠️ 低血糖风险: 低
      💡 建议: 正常晚餐即可,睡前可适当加餐

场景 9:趋势预测

用户: 预测一下明天的血糖
助手: 🔮 血糖趋势预测(基于近期数据)
      
      📅 明日预测:
      • 空腹: 6.2-6.8 mmol/L
      • 早餐后: 8.5-9.5 mmol/L
      • 午餐后: 8.0-9.0 mmol/L
      • 晚餐后: 8.5-10.0 mmol/L
      
      💡 建议: 早餐注意控制碳水,预计餐后偏高

💡 健康建议场景

场景 10:饮食建议

用户: 晚餐吃什么比较好
助手: 🍽️ 晚餐建议(基于今日血糖)
      
      📊 今日血糖回顾:
      • 空腹: 6.9 mmol/L(略高)
      • 午餐后: 9.5 mmol/L(偏高)
      
      🥗 推荐晚餐:
      • 主食: 杂粮饭 1小碗(约100g)
      • 蛋白质: 清蒸鱼/鸡胸肉
      • 蔬菜: 西兰花、菠菜等绿叶菜
      
      ⚠️ 避免: 白米饭、面条、高糖水果

场景 11:运动建议

用户: 什么时候运动比较好
助手: 🏃 运动时间建议(基于您的血糖模式)
      
      📊 最佳运动时间:
      1. 早餐后1小时 (8:00-9:00)
         → 血糖约 8.0 mmol/L,适合运动
      2. 晚餐后1小时 (19:00-20:00)
         → 血糖约 9.0 mmol/L,适合运动
      
      ⚠️ 避免时间:
      • 空腹晨练(血糖偏低风险)
      • 睡前运动(影响睡眠质量)
      
      💡 建议运动: 快走30分钟或慢跑20分钟

场景 12:异常预警

用户: 帮我分析一下最近血糖
助手: ⚠️ 血糖异常预警
      
      🔴 发现以下问题:
      1. 空腹血糖连续3天 >7.0 mmol/L
      2. 餐后血糖波动大,最高达 12.5 mmol/L
      3. 达标率仅 55%,低于目标
      
      💡 紧急建议:
      1. 建议增加监测频率,每日4次以上
      2. 严格控制碳水摄入,选择低GI食物
      3. 考虑就医咨询,调整治疗方案

技术示例

以下为开发参考的技术调用示例:

记录血糖读数

示例 1:简单录入

用户: "我血糖 7.2 mmol/L"
→ 执行: scripts/record_glucose.py --value 7.2 --unit mmol/L

示例 2:详细录入

用户: "今天早上空腹血糖 5.8,早餐前测的"
→ 执行: scripts/record_glucose.py --value 5.8 --unit mmol/L --meal-context fasting --time "2026-03-24 07:00"

导入数据

示例 3:从 CSV 导入

用户: "从 readings.csv 导入我的血糖数据"
→ 执行: scripts/import_glucose_data.py --file readings.csv --format csv

示例 4:从 Excel 导入

用户: "加载我的 Excel 血糖记录"
→ 执行: scripts/import_glucose_data.py --file glucose_log.xlsx --format excel

分析趋势

示例 5:周报告

用户: "看看我这周的血糖趋势"
→ 执行: scripts/analyze_glucose.py --period week --output chart

示例 6:模式分析

用户: "分析我的餐后血糖模式"
→ 执行: scripts/analyze_glucose.py --type post-meal --output report

预测

示例 7:短期预测

用户: "预测接下来几小时的血糖"
→ 执行: scripts/predict_glucose.py --horizon 4h

示例 8:风险评估

用户: "今晚我有低血糖风险吗?"
→ 执行: scripts/predict_glucose.py --risk-type hypoglycemia --period tonight

获取建议

示例 9:饮食建议

用户: "根据今天的血糖,晚餐该吃什么?"
→ 执行: scripts/generate_recommendations.py --type meal --meal dinner

示例 10:运动规划

用户: "今天什么时候适合运动?"
→ 执行: scripts/generate_recommendations.py --type activity

工作流程

  1. 初始设置:首次使用时,创建数据目录并初始化存储文件
  2. 数据收集:定期记录或导入血糖读数
  3. 分析:积累足够数据后运行趋势分析(建议 ≥7 天)
  4. 预测:数据积累 14 天以上后使用预测模型以获得更好准确性
  5. 建议:基于模式和预测生成个性化建议

重要说明

医疗免责声明

  • 本技能仅供信息记录和追踪用途
  • 不能替代专业医疗建议、诊断或治疗
  • 在做出治疗决策前请务必咨询医疗保健提供者
  • 预测结果为统计估计,非医疗诊断

数据质量

  • 预测准确性取决于数据质量和一致性
  • 定期、一致的读数可改善模式识别
  • 包含餐前/餐后状态和备注可获得更好的分析结果

隐私保护

  • 所有数据存储在用户本地设备
  • 不向外部服务器传输任何数据
  • 用户对自己的健康数据拥有完全控制权

资源目录

scripts/ 脚本目录

核心脚本

  • record_glucose.py - 记录手动血糖录入
  • import_glucose_data.py - 从 CSV/Excel 文件导入数据
  • analyze_glucose.py - 执行统计分析并生成图表
  • predict_glucose.py - 运行血糖预测模型
  • generate_recommendations.py - 创建个性化健康建议
  • data_manager.py - 处理数据存储、备份和检索

工具脚本

  • convert_units.py - 在 mmol/L 和 mg/dL 之间转换
  • validate_data.py - 验证血糖数据格式和数值
  • export_report.py - 导出各种格式的分析报告

references/ 参考资料

医疗参考

  • glucose_ranges.md - 正常、糖前期和糖尿病血糖水平参考范围
  • glycemic_index.md - 常见食物的血糖指数值
  • exercise_guidelines.md - 运动与血糖管理指南

技术文档

  • data_schema.md - JSON 数据结构规范
  • prediction_algorithms.md - 预测方法文档
  • analysis_methods.md - 趋势分析中使用的统计方法

assets/ 资源文件

模板

  • glucose_log_template.csv - CSV 血糖记录模板(示例数据)
  • glucose_log_template.xlsx - Excel 血糖记录模板(示例数据)
  • data_template.json - JSON 数据模板

可视化

  • chart_styles.json - 血糖图表默认样式
  • report_template.html - 分析报告 HTML 模板

Comments

Loading comments...