可信 — 风险评分 5/100
上次扫描:1 天前 重新扫描
5 /100
tracked-video-analysis
Analyze local or linked video files and convert them into structured summaries of features, functions, workflows, or topics
Legitimate video transcription skill with transparent subprocess usage for ffmpeg/ffprobe and model downloads from HuggingFace, no suspicious behavior detected.
技能名称tracked-video-analysis
分析耗时24.7s
引擎pi
可以安装
This skill is safe to use. No security concerns requiring action.
资源类型声明权限推断权限状态证据
文件系统 WRITE WRITE ✓ 一致 SKILL.md line 17: tmp/video_analysis/ working directory
网络访问 READ READ ✓ 一致 transcribe_tracked_light.mjs:28: Xenova/whisper-tiny from HuggingFace
命令执行 WRITE WRITE ✓ 一致 transcribe_tracked_light.mjs:34-35: execFileSync for ffmpeg/ffprobe
环境变量 NONE NONE No environment variable access observed
技能调用 NONE NONE No skill invocation found
剪贴板 NONE NONE No clipboard access found
浏览器 NONE NONE No browser automation found
数据库 NONE NONE No database access found

目录结构

4 文件 · 15.7 KB · 463 行
Markdown 2f · 274L Python 1f · 120L JavaScript 1f · 69L
├─ 📁 references
│ └─ 📝 pipeline.md Markdown 145L · 3.0 KB
├─ 📁 scripts
│ ├─ 🐍 final_structurer.py Python 120L · 5.3 KB
│ └─ 📜 transcribe_tracked_light.mjs JavaScript 69L · 3.5 KB
└─ 📝 SKILL.md Markdown 129L · 4.0 KB

依赖分析 4 项

包名版本来源已知漏洞备注
@xenova/transformers latest npm Legitimate HuggingFace Transformers port for Node.js
ffmpeg-static latest npm Static ffmpeg binary for video processing
ffprobe-static latest npm Static ffprobe binary for video metadata
wavefile latest npm WAV file processing library

安全亮点

✓ All subprocess calls are documented in SKILL.md
✓ Model download from HuggingFace is a legitimate AI/ML source
✓ File operations scoped to tmp/video_analysis/ as documented
✓ No credential access or environment variable harvesting
✓ No data exfiltration or C2 communication
✓ No obfuscation or base64-encoded execution
✓ Clean two-stage workflow with explicit status tracking
✓ No remote script execution (curl|bash patterns)
✓ Dependencies (@xenova/transformers, ffmpeg-static, wavefile) are standard ML/video processing tools