audiocraft-audio-generation
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation provides instructions to install the
audiocraftlibrary from Meta's official GitHub repository and includes standard research tools likedora-searchfor experiment management. - [COMMAND_EXECUTION]: The skill includes standard shell commands for environment configuration, such as installing Python packages via
pipand system-level audio utilities likeffmpegusingapt-get. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates processing user-supplied text descriptions to generate audio content. While the examples do not use explicit delimiters or sanitization for these inputs, the potential for system-level impact is minimal given the specialized nature of the audio generation models.
- Ingestion points: User-provided text prompts entered via Gradio web interfaces, FastAPI service endpoints, or raw string inputs in Python scripts.
- Boundary markers: Not present in the provided code snippets.
- Capability inventory: Audio generation model inference and saving resulting waveforms to the local filesystem.
- Sanitization: No specific text filtering or prompt sanitization logic is demonstrated in the code examples.
Audit Metadata