ai-avatar-video
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions use the
infshCLI tool for video generation. Usage is properly scoped via theallowed-toolsproperty in the manifest. - [EXTERNAL_DOWNLOADS]: The skill provides links to official
inference.shdocumentation and related utility skills on GitHub. These resources are relevant to the primary function and come from the service provider's repository. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes user-provided media URLs. 1. Ingestion points:
image_urlandaudio_urlparameters inSKILL.md. 2. Boundary markers: Inputs are contained within JSON structures. 3. Capability inventory: The skill is limited toinfshtool calls. 4. Sanitization: Relies on the external CLI's input handling.
Audit Metadata