musia-music-production

Pass

Audited by Gen Agent Trust Hub on Oct 19, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to interact with local tools such as node bin/musia.js and various Python scripts. These commands are part of the intended workflow and are executed within the user's local project environment.
  • [INDIRECT_PROMPT_INJECTION]: The agent is instructed to process external lyrics and classical poems. While this ingestion represents a theoretical attack surface for indirect prompt injection, the skill's focus on creative production and the requirement for manual human review and audio alignment significantly mitigate this risk.
  • [EXTERNAL_DOWNLOADS]: The instructions mention the possibility of downloading newer model checkpoints if available from upstream projects. This is presented as a manual or user-requested step for quality improvement rather than an automated execution of unverified remote code.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 19, 2026, 06:48 PM