Skill flagged — suspicious patterns detected

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

Cinmoore Skill

v1.0.0

神眸品牌摄像头技能:相机PTZ控制;图像/视频数据采集;多模态大模型图像/视频分析、图像编辑、智能寻物定位;飞书推送;自定义检测事件;基于视频分析的自动vlog生成。使用场景:(1) 控制相机移动和录制 (2) 抓取图像和视频 (3) 分析图像/视频内容 (4) 智能寻物定位 (5) 图像编辑 (6) 发送消息到...

0· 76·0 current·1 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 buaalingming/cinmoore-skill.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Cinmoore Skill" (buaalingming/cinmoore-skill) from ClawHub.
Skill page: https://clawhub.ai/buaalingming/cinmoore-skill
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Required binaries: python
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 cinmoore-skill

ClawHub CLI

Package manager switcher

npx clawhub@latest install cinmoore-skill
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The SKILL.md functionality (PTZ control, capture, AI analysis, Feishu push, vlog generation) matches the skill name/description. However, the skill expects a local configuration file containing Feishu credentials and a DashScope API key (for models), but the registry metadata lists no required env vars or credentials — an inconsistency between claimed needs and declared requirements.
!
Instruction Scope
Runtime instructions instruct the agent/user to run python modules (python -m cinmoore_skill.*) and to create a cinmoore_config.yaml containing device_id, feishu app_id/app_secret, and models.api_key. The SKILL.md also directs reading and writing local files under temp\ and pushing data to external services (Feishu and DashScope). The instructions reference modules and files that are not present in the skill bundle, giving the agent/user broad discretion to install/run external code.
!
Install Mechanism
The registry includes an install command that runs pip install on a relative wheel path ('python -m pip install wheel\cinmoore_skill-1.0.0-py3-none-any.whl') on Windows. No wheel file is bundled with the skill. This is suspicious because it implies relying on an external/local wheel (not from a known registry or release host); if executed, pip-installing an unknown wheel would run arbitrary code on the host.
!
Credentials
The registry declares no required environment variables or primary credential, but the SKILL.md requires a config file with Feishu app_id/app_secret and a DashScope API key. Secrets are expected to be stored in a local YAML rather than declared in the skill manifest, which hides needed privileges and is disproportionate to the declared metadata.
Persistence & Privilege
The skill does not request always:true, does not claim system-wide persistence, and is user-invocable only. It does instruct storing outputs and configs in its own directory (cinmoore-skill and temp\), which is normal for a local camera integration.
What to consider before installing
This skill appears to need Feishu credentials and a DashScope API key (sensitive secrets) and expects you to install a wheel that is not bundled with the skill. Before installing or using it: (1) Ask the publisher for the actual wheel/source code and verify its origin and contents (check a signed release or GitHub release) rather than allowing an arbitrary pip install; (2) Do not provide Feishu app_secret or DashScope API keys until you verify the code and trust the publisher; (3) Prefer running first in an isolated/sandboxed environment or VM; (4) If you must test, inspect the wheel or source for network calls, credential exfiltration, or unexpected system access; (5) Request the owner to update the skill manifest to declare required credentials and provide a safe install method (official PyPI/GitHub release). These discrepancies (missing declared credentials and a referenced-but-missing wheel) are the main red flags.

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

Runtime requirements

🖼️ Clawdis
Binspython
latestvk976eat52zsbj1df12ghjf61qd842g0b
76downloads
0stars
1versions
Updated 3w ago
v1.0.0
MIT-0

神眸 (Cinmoore) 技能指南

智能摄像头控制、媒体采集、AI视觉分析与飞书消息推送一体化解决方案。

核心能力

  • 📷 相机控制:PTZ云台控制、视频录制、图像抓取
  • 🎨 媒体分析:图像/视频AI分析、智能寻物定位、图像编辑转换
  • 📬 飞书推送:图片/视频卡片消息、自动GIF预览
  • 🎬 Vlog生成:自动视频剪辑与合成
  • ⚠️ 事件检测:自定义事件监控与告警

配置说明

⚠️ 重要:首次使用前必须在cinmoore-skill目录下配置 cinmoore_config.yaml 文件, 同时所有命令需要在cinmoore-skill目录下执行,以确保配置文件准确加载,数据统一保存在该目录下。

使用提醒

⚠️ 重要:PowerShell 不支持 && 连接多条命令。如需顺序执行多条命令:

  1. 使用分号 ; 分隔:cmd1; cmd2
  2. 或使用管道 | 传递输出:cmd1 | cmd2

初始化配置文件:

# 在cinmoore-skill目录下执行,然后提示用户提供必要配置项
python -m cinmoore_skill.commands.init

Windows编码设置(避免中文乱码):

chcp 65001
$env:PYTHONIOENCODING="utf-8"

必需配置项:

server:
  device_id: "YOUR_DEVICE_ID"  # 相机设备ID

feishu:
  app_id: "YOUR_APP_ID"
  app_secret: "YOUR_APP_SECRET"
  receive_id: "YOUR_RECEIVE_ID"

models:
  api_key: "YOUR_API_KEY"  # 阿里云 DashScope API Key

快速开始

1. 相机控制

# 抓取图像(录制2秒视频并提取中间帧,不会引起云台转动)
python -m cinmoore_skill.commands.test_camera_controller --capture

# 录制5秒视频(默认不转动云台)
python -m cinmoore_skill.commands.test_camera_controller --record-duration 5.0

# 只转动云台不录制视频(向右移动)
python -m cinmoore_skill.commands.test_camera_controller --ptz-only --ptz-direction 1 --ptz-speed 0 --ptz-duration 3.0

# 转动云台并向右移动(会自动录制视频)
python -m cinmoore_skill.commands.test_camera_controller --ptz --ptz-direction 1 --ptz-speed 0 --ptz-duration 3.0

参数说明:

  • --capture: 抓取单张图像(默认录制2秒视频提取中间帧)
  • --capture-duration: 抓图时录制视频时长(秒)
  • --record-duration: 录制视频时长(秒),默认不转动云台
  • --ptz: 启用PTZ云台转动并录制视频(转动+录制)
  • --ptz-only: 只转动PTZ云台不录制视频(仅转动)
  • --ptz-direction: 运动方向(0:左, 1:右, 2:上, 3:下)
  • --ptz-speed: 运动速度(0:慢, 1:中, 2:快)
  • --ptz-duration: 运动持续时间(秒)

逻辑说明

  • 仅使用 --record-duration:只录制视频,云台不转动
  • 使用 --ptz-only:只转动云台,不录制视频
  • 使用 --ptz:转动云台并自动录制视频

2. 媒体分析

# 分析图片内容
python -m cinmoore_skill.commands.test_media_analysis --analyze "描述图片中的场景和人物" image.jpg

# 智能寻物:定位并标注图片中的对象
python -m cinmoore_skill.commands.test_media_analysis --ground "找到所有人像和车辆" image.jpg

# 图像编辑:风格转换、内容修改
python -m cinmoore_skill.commands.test_media_analysis --edit "将背景换成海滩" image.jpg

# 分析视频内容
python -m cinmoore_skill.commands.test_media_analysis --video-analyze "分析视频中的主要事件" video.mp4

参数说明:

  • --analyze: 分析图像内容,支持场景描述、人物识别、异常检测
  • --ground: 智能寻物定位,返回带标注的图像
  • --edit: 图像编辑,支持风格转换、背景替换、内容修改
  • --video-analyze: 视频内容分析,提取关键帧并分析事件

3. 飞书推送

# 发送图片消息
python -m cinmoore_skill.commands.test_feishu_controller --image "path\to\image.jpg" --title "巡检截图"

# 发送视频消息(自动生成GIF预览)
python -m cinmoore_skill.commands.test_feishu_controller --video "path\to\video.mp4" --title "巡检录像"

# 发送消息并附带AI分析结果
python -m cinmoore_skill.commands.test_feishu_controller --image "path\to\image.jpg" --title "安全巡检" --analysis "检测到有人员闯入"

回复准则: 发送成功后,根据内容简洁回答或询问,像家人一样,不超过100字。

4. 事件检测

# 查看当前事件列表
python -m cinmoore_skill.commands.test_event_reporter set --list

# 添加检测事件
python -m cinmoore_skill.commands.test_event_reporter set --add "检测有人闯入"

# 删除检测事件
python -m cinmoore_skill.commands.test_event_reporter set --delete "检测有人闯入"

# 运行事件检测
python -m cinmoore_skill.commands.test_event_reporter run --txt_file analysis.txt

5. Vlog生成

# 基础生成(默认主题:日常生活)
python -m cinmoore_skill.commands.test_generate_vlog video_folder\

# 指定主题
python -m cinmoore_skill.commands.test_generate_vlog video_folder\ --user_input "乒乓球比赛"

# 指定时间范围
python -m cinmoore_skill.commands.test_generate_vlog video_folder\ --user_input "乒乓球比赛" --start_time 20260311000000 --end_time 20260311235959

输入说明: 程序会自动检查指定文件夹是否存在匹配的视频和描述文件,无需手动创建

  • 需要有已分析的视频文件(.mp4)和对应的分析文本文件(analysis_视频文件名.txt
  • 程序会自动查找匹配的文件进行vlog生成

输出文件:

  • vlog_{时间戳}.mp4: 合成视频
  • vlog_{时间戳}_cover.jpg: 封面图片

输出路径管理

所有输出文件保存在 temp\ 目录,按日期分目录:

temp\
├── 20260317\
│   ├── image_设备ID_时间.jpg   # 抓取图像
│   ├── video_设备ID_时间.mp4   # 录制视频
│   ├── analysis_时间.txt       # AI分析结果
│   ├── grounded_时间.jpg       # 寻物标注
│   ├── edited_时间.jpg         # 图像编辑
│   ├── gif_时间.gif            # GIF预览
│   └── vlog_时间.mp4           # Vlog视频
├── custom_event.json           # 事件定义

获取最新文件:

$IMAGE = Get-ChildItem -Path "temp\*\image_*.jpg" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }
$VIDEO = Get-ChildItem -Path "temp\*\video_*.mp4" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }
$ANALYSIS = Get-ChildItem -Path "temp\*\analysis_*.txt" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }

工作流示例

安全巡检工作流

Write-Host "=== 安全巡检工作流 ==="

# 步骤1:抓取巡检图像
python -m cinmoore_skill.commands.test_camera_controller --capture --capture-duration 3
$IMAGE_FILE = Get-ChildItem -Path "temp\*\image_*.jpg" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }
Write-Host "✓ 图像已保存: $IMAGE_FILE"

# 步骤2:AI分析图像内容
python -m cinmoore_skill.commands.test_media_analysis --analyze "分析巡检情况,识别异常" $IMAGE_FILE
$ANALYSIS_FILE = Get-ChildItem -Path "temp\*\analysis_*.txt" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }
Write-Host "✓ 分析报告已保存: $ANALYSIS_FILE"

# 步骤3:读取分析结果并推送到飞书
$ANALYSIS_CONTENT = Get-Content $ANALYSIS_FILE
python -m cinmoore_skill.commands.test_feishu_controller --image $IMAGE_FILE --title "安全巡检报告" --analysis $ANALYSIS_CONTENT
Write-Host "✓ 已推送到飞书"

Write-Host "=== 工作流完成 ==="

智能寻物工作流

Write-Host "=== 智能寻物工作流 ==="

# 步骤1:抓取图像
python -m cinmoore_skill.commands.test_camera_controller --capture
$IMAGE_FILE = Get-ChildItem -Path "temp\*\image_*.jpg" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }

# 步骤2:定位并标注目标
python -m cinmoore_skill.commands.test_media_analysis --ground "找到红色背包" $IMAGE_FILE
$GROUNDED_FILE = Get-ChildItem -Path "temp\*\grounded_*.jpg" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }

# 步骤3:发送标注结果
python -m cinmoore_skill.commands.test_feishu_controller --image $GROUNDED_FILE --title "寻物结果"

Write-Host "=== 工作流完成 ==="

事件监控工作流

Write-Host "=== 事件监控工作流 ==="

# 步骤1:设置检测事件(首次运行)
python -m cinmoore_skill.commands.test_event_reporter set --add "检测有人闯入" --add "检测到烟雾"

# 步骤2:抓取并分析
python -m cinmoore_skill.commands.test_camera_controller --capture
$IMAGE_FILE = Get-ChildItem -Path "temp\*\image_*.jpg" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }

python -m cinmoore_skill.commands.test_media_analysis --analyze "分析场景中的异常情况" $IMAGE_FILE
$ANALYSIS_FILE = Get-ChildItem -Path "temp\*\analysis_*.txt" | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | ForEach-Object { $_.FullName }

# 步骤3:运行事件检测(自动推送告警)
python -m cinmoore_skill.commands.test_event_reporter run --txt_file $ANALYSIS_FILE

Write-Host "=== 工作流完成 ==="

Comments

Loading comments...