Skills
Navigate
skills/scenario-labs/skills/scenario-seedance-music-video/Gen Agent Trust Hub

scenario-seedance-music-video

Pass

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes Python scripts scripts/build.py and scripts/song.py that execute ffmpeg and ffprobe for media assembly and analysis. These tools are invoked using subprocess.run with argument lists, which prevents shell injection by avoiding shell interpretation of inputs.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied audio files and JSON edit metadata, creating an attack surface for indirect instructions. This is mitigated by input validation and structured parsing. Evidence chain:\n
  • Ingestion points: Audio files (master.mp3) and structured edit data (edit.json) processed by scripts/build.py and scripts/song.py.\n
  • Boundary markers: None.\n
  • Capability inventory: Media processing via subprocess.run calls to ffmpeg and ffprobe.\n
  • Sanitization: Employs Path.resolve() for file paths and performs range validation for numeric parameters such as audio gain and frame rates.\n- [EXTERNAL_DOWNLOADS]: The documentation references fetching generated media assets via curl from platform-provided URLs, which is standard behavior for managing project resources.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 17, 2026, 09:25 PM