ClawHub
Back to skill

Security audit

cutmv

Security checks across malware telemetry and agentic risk

Overview

This is a legitimate local video-editing tool, but it needs review because it evaluates media metadata as Python code and overwrites output files by default.

Install only if you are comfortable reviewing or patching the eval() call first. Avoid using the info command on untrusted media, use a trusted FFmpeg/ffprobe installation, and choose output paths carefully because existing files may be overwritten without confirmation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

eval() call detected

High
Category
Dangerous Code Execution
Content
"codec": video_stream.get("codec_name") if video_stream else None,
                "width": video_stream.get("width") if video_stream else None,
                "height": video_stream.get("height") if video_stream else None,
                "fps": eval(video_stream.get("r_frame_rate", "0/1")) if video_stream else None,
            } if video_stream else None,
            "audio": {
                "codec": audio_stream.get("codec_name") if audio_stream else None,
Confidence
99% confidence
Finding
The code evaluates ffprobe-derived data with Python eval(), which can execute arbitrary Python code if the parsed value is attacker-controlled or if a malicious ffprobe binary/output is introduced. In this skill context, media metadata comes from untrusted files and external tooling, so treating it as executable code is unsafe and can lead to code execution in the process running the skill.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The helper forces ffmpeg to run with '-y', automatically overwriting output files without confirmation. In an agent/skill context where file paths may be derived from user or upstream workflow input, this can cause unintended data destruction or clobber important files, making the behavior more dangerous than in a purely interactive local CLI.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.dynamic_code_execution

Dynamic code execution detected.

Critical
Code
suspicious.dynamic_code_execution
Location
skill.py:319