media-processing

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • Command Execution (SAFE): Analysis of scripts/media_convert.py shows that external binaries are invoked using subprocess.run() with argument lists instead of shell strings. This design choice effectively mitigates shell injection vulnerabilities that could otherwise occur if malicious filenames were processed.
  • Data Exposure & Exfiltration (SAFE): The skill does not attempt to access sensitive file paths (such as SSH keys or environment files) and contains no hardcoded credentials. Network operations described in the documentation, such as RTMP streaming, are consistent with the skill's primary media-processing purpose.
  • Dependency Security (SAFE): The skill relies on standard system binaries (FFmpeg, ImageMagick) and reputable Python testing libraries (pytest). No suspicious or unverifiable third-party packages are included.
  • Indirect Prompt Injection (SAFE): While the skill ingests media files in media_convert.py (ingestion point), it treats them as binary data for processing via ffmpeg (capability inventory). It does not parse media content for instructions, and filenames are sanitized via list-based argument passing (sanitization), preventing injection into the agent's logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:41 PM