ai-music-audio
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The audio watermarking implementation in
references/patterns.mduseschild_process.spawnto run an embedded Python script. This represents a dynamic execution surface used for cross-language interoperability, implemented safely without using a shell. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input for music and speech generation. Ingestion points: Prompt and text fields in
references/patterns.md. Boundary markers: None in implementation patterns; recommended inreferences/sharp_edges.md. Capability inventory: Subprocess execution and network requests inreferences/patterns.md. Sanitization: Absent in base code; remediation provided inreferences/sharp_edges.md. - [COMMAND_EXECUTION]: The skill uses
child_process.spawnto interface with specialized Python audio tools. The implementation avoids shell injection by passing arguments as an array rather than a single string.
Audit Metadata