skywork-music-maker

Pass

Audited by Gen Agent Trust Hub on Mar 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by transforming untrusted natural language music descriptions into structured prompts for the Mureka AI API.
  • Ingestion points: Natural language descriptions and lyrics provided by the user (as detailed in SKILL.md and scripts/mureka.py).
  • Boundary markers: The skill uses a 'Smart Prompt Conversion' workflow and 'Quality Checklist' to structure and validate inputs, though it lacks explicit technical delimiter markers in the final API payload construction.
  • Capability inventory: Network communication with api.mureka.ai and file-write capabilities for saving audio and lyrics files via scripts/mureka.py.
  • Sanitization: The risk is significantly mitigated by a mandatory human-in-the-loop confirmation step where the generated prompt must be presented to the user for approval before the generation task is submitted.
  • [SAFE]: The skill correctly handles secrets by utilizing environment variables (MUREKA_API_KEY) rather than hardcoded credentials.
  • [SAFE]: All network operations are restricted to the vendor's official domain (api.mureka.ai), and external downloads are limited to the resulting media files generated by the service.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 19, 2026, 02:39 AM