stable-audio
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted audio files (mp3, wav) provided by users for audio-to-audio and inpainting workflows, which constitutes an indirect prompt injection surface.
- Ingestion points: User-uploaded audio files as described in the 'Capability map' and 'Workflow patterns' sections of SKILL.md.
- Boundary markers: The instructions rely on external provider scanning (Stability AI) and manual rights-clearance checks but do not implement specific agent-side prompt delimiters for data processing.
- Capability inventory: The skill uses network capabilities to interact with the Stability API and performs file-system operations to manage audio assets.
- Sanitization: There is no mention of sanitizing file metadata or content at the agent level.
- [SAFE]: The skill references official Stability AI domains and repositories for API access and documentation, representing interactions with a well-known technology service provider.
Audit Metadata