audio-denoise-api
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEMETADATA_POISONINGEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill metadata contains conflicting instructions; the description and trigger logic refer to 'image generation' and explicitly prohibit triggering for 'voice' or 'audio', which contradicts the core functionality, parameters, and API endpoints defined for audio denoising.
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to retrieve extended documentation and parameter schemas from the vendor's official domain at
https://www.pixazo.ai/models/audio-denoise-api.mdto supplement its execution logic. - [COMMAND_EXECUTION]: The instructions guide the agent to perform local file system operations, including creating a configuration directory and applying restrictive permissions (
chmod 600) to secure the API key file. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data via user-provided prompts and the
audio_urlparameter, which could contain instructions intended to influence the agent's behavior. - Ingestion points: User-supplied text prompts and
audio_urlfield in the API request. - Boundary markers: None identified in the skill instructions.
- Capability inventory: Network requests (API gateway) and local file writing for credential storage.
- Sanitization: No validation or filtering logic is specified for input parameters or API responses.
Audit Metadata