Skill flagged — suspicious patterns detected

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

Daily Meal Planner

v1.2.6

每日智能菜谱推荐。触发词:今天吃什么/中午吃什么/晚餐推荐/下午茶/夜宵/一周菜单/清淡/辣的/快手菜/减肥。支持按餐次、口味、心情、季节、天气、地域智能推荐,带详细做法和营养数据。

0· 198·0 current·0 all-time
byJiaqi@gmmg55

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for gmmg55/daily-meal-planner.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Daily Meal Planner" (gmmg55/daily-meal-planner) from ClawHub.
Skill page: https://clawhub.ai/gmmg55/daily-meal-planner
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 daily-meal-planner

ClawHub CLI

Package manager switcher

npx clawhub@latest install daily-meal-planner
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name/description align with included code: scripts implement recommendation, local profile, weather lookup and remote JSON data for meals. The declared dependencies (none required at runtime; requirements.txt lists requests) and declared network calls (GitHub/jsdelivr and wttr.in) are coherent with the stated functionality.
!
Instruction Scope
SKILL.md instructs the agent to download recipe DB from GitHub/jsdelivr and to query wttr.in (both implemented). However the doc contains a detected 'unicode-control-chars' prompt‑injection signal and asks the AI to 'decide itself' where to get holidays (vague, grants broad discretion). The instructions instruct writing a local user_profile.json (expected) but also contain free-form template/behavior rules that could influence model outputs and potentially be abused if malicious content is embedded.
Install Mechanism
No install spec; files are shipped with the skill. The only package hint is scripts/requirements.txt (requests>=2.28.0). No archive downloads or remote installers are used by the installer; runtime code downloads data files from GitHub/jsdelivr (expected for this skill).
Credentials
The skill does not request environment variables, credentials, or access to unrelated config paths. It stores a local user_profile.json and downloaded JSON under the scripts directory — consistent with the stated purpose. Network access to public endpoints (GitHub/jsdelivr, wttr.in) is required and proportional to functionality.
Persistence & Privilege
always:false and no special platform privileges. The skill writes a local user_profile.json (explained in docs). It does not request to modify other skills or global agent settings.
Scan Findings in Context
[unicode-control-chars] unexpected: Control/unicode control characters in SKILL.md are unusual for recipe text and are a prompt-injection indicator; they could be accidental encoding artefacts but merit manual inspection and removal before trusting the instructions.
What to consider before installing
This skill appears to do what it says (meal recommendations, local profile, weather-aware suggestions), but take precautions before installing: 1) Inspect or vet the referenced GitHub repo (https://github.com/GMMG55/daily-meal-planner) and the downloaded JSON files — the runtime downloads recipe data from GitHub/jsdelivr. 2) Remove or examine any hidden/control characters found in SKILL.md (scan found unicode-control-chars). 3) Run first in a sandbox/container or with limited network access to confirm behavior. 4) If you rely on holiday data, note the SKILL.md asks the AI to choose sources (this is vague); prefer explicit trusted endpoints. 5) If you are uncomfortable with automatic downloads, manually prefetch and review the JSON files and remove network calls in the scripts. If you want higher assurance, ask the author for a pinned release tarball or signed release instead of raw branch downloads.

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

latestvk975n9h8wh12na4swds9k8etf185pbdk
198downloads
0stars
17versions
Updated 1h ago
v1.2.6
MIT-0

Daily Meal Planner

智能推荐菜谱,支持用户画像记忆和偏好学习。每餐推荐3套完整套餐(主菜+配菜+汤/主食)。

⚠️ 安全与隐私说明

网络请求

  • 首次运行时会从 GitHub/jsdelivr(官方镜像)下载菜谱数据库
  • 查询 wttr.in 获取实时天气
  • 查询 date.nager.at 获取节假日信息
  • 不使用第三方代理(已移除 ghproxy)

数据存储

  • 会在本地保存 user_profile.json(用户画像,包含地点、口味偏好等)
  • 如不希望持久化,可删除该文件
  • 下载的 JSON 数据文件存储在 scripts 目录

无敏感数据

  • 不请求任何密钥、系统凭据或个人隐私信息
  • 不收集或上传用户数据

安全建议

⚠️ AI 执行规则

本 skill 不需要保存任务摘要文件。直接在对话中展示推荐结果即可,无需写入任何 md 文件。

🎯 推荐算法优化规则

1. 餐次权重优先+大模型智能推荐

示例

  • 早餐:优先推荐快手菜(15分钟内完成)、清淡养胃、粥类、营养均衡
    • ❌ 避免推荐:硬菜、大荤、功夫菜(如开水白菜)、耗时>30分钟的菜
    • ✅ 优先推荐:蒸蛋、粥、三明治、煎蛋、快手面食、牛奶燕麦等
  • 午餐:推荐饱腹、家常、营养均衡
  • 晚餐:推荐清淡养胃、快手、可以硬菜犒劳自己
  • 夜宵:推荐清淡、快手、粥类、小食、炸串、小啤酒

2. 使用大模型智能推荐

不要死板地使用固定模式,要根据用户画像、天气、心情、时间段综合判断: 示例

  • 用户说"疲惫" → 推荐清淡养胃、快手菜
  • 用户说"想吃川菜" → 优先川菜,但早餐要注意不要太辣
  • 用户说"减肥" → 推荐低卡、高蛋白
  • 下雨天 → 推荐热汤、暖身菜
  • 天气热 → 推荐清淡、凉菜

3. 推荐理由优化

每道菜的推荐理由要:

  • 详细具体:不要只说"周二营养跟上",要说明为什么推荐这道菜
  • 可爱有食欲:使用表情符号、口语化表达
  • 个性化:结合用户画像、天气、心情

示例

  • ✅ "早餐吃蒸蛋羹,嫩滑如豆腐,温和不刺激,胃暖暖的开始新的一天~"
  • ✅ "下雨天来碗热汤,从胃暖到心,舒服~"
  • ✅ "快手早餐,10分钟搞定,懒人福音,多睡会儿也能吃好!"
  • ❌ "周二营养跟上"(太简单)

🌟 头部标题优化规则

每次推荐时,头部标题要包含以下信息(动态生成):

必须包含:

  1. 日期信息:周几
  2. 节假日提醒:如果临近节假日,显示倒计时
  3. 天气信息:查询实时天气并描述
  4. 温馨激励文案:结合周几、天气、餐次

模板示例:

早餐模板

🍽️ 📍北京 今日早餐推荐  晴天 20°C · 周三 · 春季
🌸 春天来啦!肝气旺盛的日子,来点绿叶菜疏通疏通~

明天是周四哦,还有2天就是五一假期了,天气不错,来吃点清淡营养的开启元气满满的一天吧!

晚餐模板

🍽️ 📍北京 今日晚餐推荐  雨天 15°C · 周二 · 春季
🌸 春天来啦!

周二新开始,继续加油~外面下着雨,来碗热汤暖暖身子,犒劳忙碌一天的自己!

节假日模板

🍽️ 📍北京 今日晚餐推荐  晴天 · 周五 · 春季

🎉 五一假期快乐!放假就要好好吃~来点硬菜犒劳自己!

周五模板

周五啦!周末倒计时开始!来点好吃的庆祝一下~

周中模板

这周已经过去3天了,哈哈哈哈哈哈,来点好吃的,准备迎接周末吧!

节假日与特殊日子获取

必须实时获取,不要写死!

AI 需要自己决定从哪里获取以下信息:

  1. 中国法定节假日:元旦、春节、清明、五一、端午、中秋、国庆等
  2. 世界节日/特殊日子:世界舞蹈日(4.29)、世界读书日(4.23)、国际妇女节(3.8)、世界卫生日(4.7)、世界环境日(6.5)、国际瑜伽日(6.21)、世界粮食日(10.16)、儿童节(6.1)、情人节(2.14)、圣诞节(12.25)等

特殊日子推荐权重示例

  • 世界舞蹈日 → 推荐高蛋白、有活力的菜(牛排、海鲜、意面)
  • 世界读书日 → 推荐适合看书时吃的菜(下午茶、甜点、小食)
  • 国际妇女节 → 推荐养颜美容、滋补的菜
  • 世界卫生日 → 推荐健康营养、清淡养生的菜
  • 世界环境日 → 推荐素食、环保食材
  • 国际瑜伽日 → 推荐轻食、沙拉、低卡菜
  • 世界海洋日 → 推荐海鲜类菜品

特殊日子的趣味文案:根据节日性质生成相应的温馨文案,增加趣味性和仪式感

智能三菜推荐

  1. 📌 综合智能推荐 — 季节+周几+天气+用户画像加权
  2. 🌿 时令之选 — 按当季食材筛选
  3. 🎲 随机惊喜 — 换换口味

用户画像

自动保存到本地 user_profile.json:地点、心情、想吃类别、偏好菜系、喜欢的菜、不喜欢的菜、饮食目标、过敏食材。

天气适配

11种天气类型自动适配(炎热→清热、雨天→暖汤、雾霾→护肺等)。天气来源:wttr.in实时查询 > CLI参数 > 本地保存。每次运行优先获取实时天气,失败则回退用户画像。

使用方法

python scripts/meal_recommend.py daily                    # 今日推荐
python scripts/meal_recommend.py daily -m 晚餐            # 指定餐次
python scripts/meal_recommend.py daily -w rainy           # 指定天气
python scripts/meal_recommend.py search 辣                # 搜索
python scripts/meal_recommend.py detail 西红柿牛腩         # 详细做法
python scripts/meal_recommend.py weekly                   # 一周菜单

触发场景

  • 餐次:「今天吃什么」「晚餐推荐」「下午茶」「夜宵」「一周菜单」
  • 口味:「清淡的」「辣的」「快手菜」「硬菜」
  • 营养:「低卡」「减肥」「高蛋白」「养生」「滋补」

输出格式

  ────────────────────────────────
  ☀️ 推荐1  西红柿牛腩
  📝 酸甜浓郁肉烂汤鲜
  💡 应季食材,周五庆祝,晴天清爽
  🔥 650kcal  ⏱ 30min  难度: 中等
  📊 蛋白质:28g | 铁:4mg
  🏠 南方,江浙,粤菜
  🥗 食材: 牛腩400g,番茄3个,土豆1个,洋葱半个
  🥬 +配: 清炒时蔬(120kcal)
  🍲 +汤/饭: 白灼虾(150kcal)
  ────────────────────────────────
  💡 回复「要」或「1/2/3」查看详细做法 👨‍🍳  ·  「换个」换一批

心情→菜品

疲惫→清淡养胃、忙碌→快手清淡、开心→硬菜家常、放松→素食清淡、庆祝→硬菜经典、慵懒→快手饱腹

数据文件

文件用途
meals_db_compressed.json + meals_tags_index.json42道完整菜谱(含做法、营养)
menu_names_compressed.json + tags_index.json595道菜单名索引
user_profile.json用户画像(自动生成)

数据文件首次运行时自动从 GitHub 下载。

脚本

  • scripts/meal_recommend.py — 推荐引擎
  • scripts/meal_search.py — 全网搜索提示
  • scripts/requirements.txt — 依赖

Comments

Loading comments...