mage-music-video
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run the 'ffprobe' utility to determine audio duration. This involves executing a shell command with a file path, which could be exploited for command injection if the agent does not properly sanitize filenames.
- [DYNAMIC_EXECUTION]: The use of 'ffprobe' via subprocess call to analyze local files constitutes a dynamic execution pattern, specifically a shell command invocation.
- [INDIRECT_PROMPT_INJECTION]: The skill takes user input for video 'Direction' and inserts it directly into the prompt sent to the AI model. 1. Ingestion points: User-provided text instructions and URLs for audio and image files. 2. Boundary markers: The instruction uses 'Also follow these custom instructions:' as a separator. 3. Capability inventory: Invokes external API tools (generate, estimate_cost) and local shell commands (ffprobe). 4. Sanitization: No explicit validation or filtering of user-provided text or media URLs is mentioned in the skill instructions.
Audit Metadata