p-video
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill encourages the installation of the belt CLI and additional specialized skills from the inference-sh GitHub repository and NPM ecosystem.\n- [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute belt commands for user authentication and model inference, which is the core functionality of the extension.\n- [INDIRECT_PROMPT_INJECTION]: The skill acts as a proxy for user-supplied prompts, images, and audio files sent to remote video generation models.\n
- Ingestion points: Prompt, image, and audio parameters within the JSON input for the belt app run command in SKILL.md.\n
- Boundary markers: JSON formatting is used to structure the input arguments.\n
- Capability inventory: Subprocess execution of the belt CLI via the Bash tool.\n
- Sanitization: The skill does not implement local validation or sanitization, relying instead on the safety filters of the downstream inference service.
Audit Metadata