ffmpeg-vapoursynth

Fail

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The helper script scripts/vspipe.py generates temporary Python scripts (.vpy) to be executed by the VapourSynth engine. The source_snippet function (lines 94-107) uses f-strings to embed the user-supplied input_path directly into the generated code (e.g., f'clip = core.ffms2.Source(r"{input_path}")'). This allows an attacker to terminate the string and inject arbitrary Python commands using a crafted file path (e.g., path/to/video.mkv")\nimport os;os.system("id")#).
  • [COMMAND_EXECUTION]: The skill uses subprocess.call and subprocess.Popen in scripts/vspipe.py (functions run_cmd and run_pipe) to execute system commands like vspipe and ffmpeg. While these commands are invoked using lists rather than raw shells, the script accepts arbitrary extra arguments via the --ffmpeg-extra flag (using nargs=argparse.REMAINDER), which could be misused if the AI agent passes untrusted input into these parameters.
  • [DYNAMIC_EXECUTION]: The core functionality of the skill involves the dynamic generation and execution of Python scripts. As noted in the SKILL.md documentation, .vpy files are full Python scripts that execute in the user's environment, representing a significant attack surface if the script generation logic is flawed or if the agent is instructed to run external scripts.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 27, 2026, 09:53 AM
Security Audit — agent-trust-hub — ffmpeg-vapoursynth