Skill flagged — suspicious patterns detected

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

Smart Feeder Pet Detection & Recognition Skill | 智能喂食器宠物检测识别技能

v1.0.0

Based on computer vision, automatically detects and recognizes cats and dogs appearing in the target area from the perspective of feeder/IPC cameras, and sup...

0· 65·0 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 smyx-sunjinhui/smyx-pet-detection-feeder-analysis.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Smart Feeder Pet Detection & Recognition Skill | 智能喂食器宠物检测识别技能" (smyx-sunjinhui/smyx-pet-detection-feeder-analysis) from ClawHub.
Skill page: https://clawhub.ai/smyx-sunjinhui/smyx-pet-detection-feeder-analysis
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 smyx-pet-detection-feeder-analysis

ClawHub CLI

Package manager switcher

npx clawhub@latest install smyx-pet-detection-feeder-analysis
Security Scan
Capability signals
Requires sensitive credentials
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The code and SKILL.md implement pet detection/identity enrollment and history listing as advertised. However the package also bundles a distinct 'face_analysis' skill and a sizeable smyx_common library; inclusion of that unrelated component is unusual but can be explained by shared common code. The need to read local skill/workspace config files (to obtain open-id/api-key/base URLs) is plausible for calling the remote API but is more than a minimal 'local detection' implementation would need.
!
Instruction Scope
Runtime instructions require the agent to: save uploaded attachments into a local attachments directory, read config files in the skill and workspace (skills/smyx_common/scripts/config.yaml), forcibly obtain an 'open-id' via local config or by prompting the user, and always query cloud APIs for history (python -m scripts.pet_detection_feeder_analysis --list --open-id). These steps cause local filesystem reads/writes and result in uploading user media and query parameters to remote services. The SKILL.md explicitly forbids reading local memory files and LanceDB, but allows/mandates reading config files and using the cloud API; that restriction is advisory and could be bypassed by an agent, so it is not a strong technical guarantee.
Install Mechanism
There is no install spec (instruction-only in metadata), so nothing is automatically downloaded or executed by the installer. The bundle includes Python scripts and multiple requirements.txt files (including a long smyx_common dependency list) which indicate heavy dependencies if the user runs the code. No third-party binary downloads or obscure URLs were used in the provided files.
!
Credentials
The skill declares no required environment variables, but its code reads environment variables (OPENCLAW_SENDER_OPEN_ID, OPENCLAW_SENDER_USERNAME, FEISHU_OPEN_ID) and automatically loads YAML config files (skills/smyx_common/scripts/config.yaml and environment-specific variants) that can contain API keys and base URLs. Requiring open-id from local config or user is reasonable for report association, but reading shared config files can expose other secrets (api-key, api-secret-key, base URLs) that are not documented as required in the skill metadata — this is disproportionate to a minimal pet-detection helper and could expose sensitive values.
Persistence & Privilege
always is false and the skill does not request persistent platform privileges. It writes uploaded attachments into its own attachments directory and reads config files in its workspace; that is normal for a script-driven integration and it does not modify other skills or system-wide settings in the provided code.
What to consider before installing
Before installing or running this skill: 1) Understand that media files (images/videos) and metadata will be uploaded to remote APIs (default base URLs are present in skills/smyx_common config.yaml). If you need privacy, do not upload sensitive footage. 2) Inspect skills/smyx_common/scripts/config.yaml and environment-specific config files to confirm the API endpoints and to check whether any API keys or secrets are stored there; the skill will read those configs. 3) The SKILL.md forces an 'open-id' lookup in workspace config or prompts the user — do not provide any global credentials you wouldn't share with the remote service. 4) Run this skill in an isolated/test environment first (or sandbox container) if you want to audit network traffic; review RequestUtil and api_service code to see exactly which endpoints and fields are sent. 5) If you are uncomfortable with uploading user media or exposing local config values, do not enable or run the skill. If you want a lower-risk option, prefer a purely local inference implementation (no network calls) or require explicit user confirmation before each upload.
!
skills/smyx_common/scripts/config-dev.yaml:2
Install source points to URL shortener or raw IP.
About static analysis
These patterns were detected by automated regex scanning. They may be normal for skills that integrate with external APIs. Check the VirusTotal and OpenClaw results above for context-aware analysis.

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

latestvk97cw19f0dd538zrhnjqjtf6as851rt7
65downloads
0stars
1versions
Updated 1w ago
v1.0.0
MIT-0

Smart Feeder Pet Detection & Recognition Skill | 智能喂食器宠物检测识别技能

Based on advanced computer vision and deep learning technologies, this feature automatically detects and identifies pets such as cats and dogs within a target area from the specific perspective of smart feeders or IPC cameras. The system not only supports high-precision breed determination but also possesses powerful individual identity recognition capabilities, allowing users to establish a dedicated database of pet facial or body features. In smart feeding scenarios, this function accurately distinguishes between different individuals in multi-pet households, enabling personalized "recognition-based feeding" services. This effectively prevents non-target pets from stealing food, providing reliable technical support for scientific pet ownership and refined health management.

本功能基于先进的计算机视觉与深度学习技术,能够从智能喂食器或IPC摄像头的特定视角出发,自动检测并识别目标区域内出现的猫、狗等宠物。系统不仅支持对宠物品种的高精度判定,更具备强大的个体身份识别能力,支持用户建立专属的宠物面部或体态特征底库。在智能喂养场景中,该功能能够精准区分多宠家庭中的不同个体,实现“认宠下粮”的个性化服务,有效防止非目标宠物抢食,为科学养宠与精细化健康管理提供可靠的技术支撑

⚠️ 强制记忆规则(最高优先级)

本技能明确约定:

  • 绝对禁止读取任何本地记忆文件:包括但不限于 memory/YYYY-MM-DD.mdMEMORY.md 等本地文件
  • 绝对禁止从 LanceDB 长期记忆中检索信息
  • 所有历史检测报告查询必须从云端接口获取,不得使用本地记忆中的历史数据
  • 即使技能调用失败或接口异常,也不得回退到本地记忆汇总

任务目标

  • 本 Skill 用于:通过喂食器/IPC摄像头视角的视频/图片进行宠物检测识别,获取结构化的宠物识别分析报告
  • 能力包含:宠物检测识别、猫/狗分类、宠物身份识别、宠物底库录入、历史检测记录查询
  • 触发条件:
    1. 默认触发:当用户提供喂食器/IPC摄像头视角的视频/图片 URL 或文件需要进行宠物检测时,默认触发本技能
    2. 当用户明确需要进行宠物检测、宠物身份识别、喂食器宠物识别、IPC摄像头宠物监测、宠物底库录入时,提及宠物检测、喂食器识别、宠物身份、底库录入等关键词,并且上传了视频文件或者图片文件
    3. 当用户提及以下关键词时,自动触发历史报告查询功能 :查看历史检测报告、宠物检测报告清单、检测报告列表、查询历史报告、显示所有检测报告、宠物识别历史记录,查询宠物检测分析报告
  • 自动行为:
    1. 如果用户上传了附件或者视频/图片文件,则自动保存到技能目录下 attachments
    2. ⚠️ 强制数据获取规则(次高优先级):如果用户触发任何历史报告查询关键词(如"查看所有检测报告"、" 显示所有宠物检测报告"、"查看历史报告"等),必须
      • 直接使用 python -m scripts.pet_detection_feeder_analysis --list --open-id 参数调用 API 查询云端的历史报告数据
      • 严格禁止:从本地 memory 目录读取历史会话信息、严格禁止手动汇总本地记录中的报告、严格禁止从长期记忆中提取报告
      • 必须统一从云端接口获取最新完整数据,然后以 Markdown 表格格式输出结果

前置准备

  • 依赖说明:scripts 脚本所需的依赖包及版本
    requests>=2.28.0
    

操作步骤

🔒 open-id 获取流程控制(强制执行,防止遗漏)

在执行宠物检测分析前,必须按以下优先级顺序获取 open-id:

第 1 步:【最高优先级】检查技能所在目录的配置文件(优先)
        路径:skills/smyx_common/scripts/config.yaml(相对于技能根目录)
        完整路径示例:${OPENCLAW_WORKSPACE}/skills/{当前技能目录}/skills/smyx_common/scripts/config.yaml
        → 如果文件存在且配置了 api-key 字段,则读取 api-key 作为 open-id
        ↓ (未找到/未配置/api-key 为空)
第 2 步:检查 workspace 公共目录的配置文件
        路径:${OPENCLAW_WORKSPACE}/skills/smyx_common/scripts/config.yaml
        → 如果文件存在且配置了 api-key 字段,则读取 api-key 作为 open-id
        ↓ (未找到/未配置)
第 3 步:检查用户是否在消息中明确提供了 open-id
        ↓ (未提供)
第 4 步:❗ 必须暂停执行,明确提示用户提供用户名或手机号作为 open-id

⚠️ 关键约束:

  • 禁止自行假设,自行推导,自行生成 open-id 值(如 openclaw-control-ui、default、petFeeder123 等)
  • 禁止跳过 open-id 验证直接调用 API
  • 必须在获取到有效 open-id 后才能继续执行分析
  • 如果用户拒绝提供 open-id,说明用途(用于保存和查询宠物检测报告记录),并询问是否继续

  • 标准流程:
    1. 准备媒体输入
      • 提供喂食器视角本地视频/图片文件路径或网络视频/图片 URL
      • 确保画面清晰展示喂食区域,光线充足
    2. 获取 open-id(强制执行)
      • 按上述流程控制获取 open-id
      • 如无法获取,必须提示用户提供用户名或手机号
    3. 执行宠物检测分析/底库录入
      • 调用 -m scripts.pet_detection_feeder_analysis 处理媒体文件(必须在技能根目录下运行脚本
      • 参数说明:
        • --input: 本地视频/图片文件路径(使用 multipart/form-data 方式上传)
        • --url: 网络视频/图片 URL 地址(API 服务自动下载)
        • --media-type: 媒体类型,可选值:video/image,默认 video
        • --pet-type: 宠物类型,可选值:cat/dog,默认 cat
        • --pet-id: 宠物ID/名称,用于底库录入(必填项,录入时必须提供)
        • --action: 操作类型,可选值:detect/enroll,默认 detect(detect=检测识别,enroll=底库录入)
        • --open-id: 当前用户的 open-id(必填,按上述流程获取)
        • --list: 显示宠物检测历史分析报告列表清单(可以输入起始日期参数过滤数据范围)
        • --api-key: API 访问密钥(可选)
        • --api-url: API 服务地址(可选,使用默认值)
        • --detail: 输出详细程度(basic/standard/json,默认 json)
        • --output: 结果输出文件路径(可选)
    4. 查看分析结果
      • 检测识别:接收结构化的宠物识别报告,包含:宠物基本信息、宠物类型、身份识别结果、置信度、出现次数统计
      • 底库录入:接收录入结果反馈,确认宠物信息已存入底库

资源索引

  • 必要脚本:见 scripts/pet_detection_feeder_analysis.py(用途:调用 API 进行宠物检测识别,本地文件使用 multipart/form-data 方式上传,网络 URL 由 API 服务自动下载)
  • 配置文件:见 scripts/config.py(用途:配置 API 地址、默认参数和媒体格式限制)
  • 领域参考:见 references/api_doc.md(何时读取:需要了解 API 接口详细规范和错误码时)

注意事项

  • 仅在需要时读取参考文档,保持上下文简洁
  • 支持格式:视频支持 mp4/avi/mov 格式,图片支持 jpg/png/jpeg 格式,最大 100MB
  • 适用于喂食器、IPC摄像头等固定视角场景,检测准确率更高
  • API 密钥可选,如果通过参数传入则必须确保调用鉴权成功,否则忽略鉴权
  • 分析结果仅供智能喂养参考,不能替代人工确认
  • 禁止临时生成脚本,只能用技能本身的脚本
  • 传入的网络地址参数,不需要下载本地,默认地址都是公网地址,api 服务会自动下载
  • 当显示历史检测报告清单的时候,从数据 json 中提取字段 reportImageUrl 作为超链接地址,使用 Markdown 表格格式输出,包含" 报告名称"、"媒体类型"、"检测时间"、"点击查看"四列,其中"报告名称"列使用宠物喂食器检测分析报告-{记录id}形式拼接, " 点击查看"列使用 [🔗 查看报告](reportImageUrl) 格式的超链接,用户点击即可直接跳转到对应的完整报告页面。
  • 表格输出示例:
    报告名称媒体类型检测时间点击查看
    宠物喂食器检测分析报告-20260312172200001视频2026-03-12 17:22:
    00🔗 查看报告

使用示例

# 检测本地视频(以下只是示例,禁止直接使用openclaw-control-ui 作为 open-id)
python -m scripts.pet_detection_feeder_analysis --input /path/to/video.mp4 --media-type video --pet-type cat --open-id openclaw-control-ui

# 检测网络视频(以下只是示例,禁止直接使用openclaw-control-ui 作为 open-id)
python -m scripts.pet_detection_feeder_analysis --url https://example.com/video.mp4 --media-type video --pet-type cat --open-id openclaw-control-ui

# 检测本地图片(以下只是示例,禁止直接使用openclaw-control-ui 作为 open-id)
python -m scripts.pet_detection_feeder_analysis --input /path/to/image.jpg --media-type image --pet-type dog --open-id openclaw-control-ui

# 宠物底库录入(将猫咪橘橘录入到底库,OpenClaw UI 上下文)
python -m scripts.pet_detection_feeder_analysis --input /path/to/juju.jpg --media-type image --pet-type cat --pet-id 橘橘 --action enroll --open-id openclaw-control-ui

# 显示历史检测报告/显示检测报告清单列表/显示历史宠物检测报告(自动触发关键词:查看历史检测报告、历史报告、检测报告清单等)
python -m scripts.pet_detection_feeder_analysis --list --open-id openclaw-control-ui

# 输出精简报告
python -m scripts.pet_detection_feeder_analysis --input video.mp4 --media-type video --pet-type cat --open-id your-open-id --detail basic

# 保存结果到文件
python -m scripts.pet_detection_feeder_analysis --input video.mp4 --media-type video --pet-type cat --open-id your-open-id --output result.json

Comments

Loading comments...