ai-video-generation
Pass
Audited by Gen Agent Trust Hub on May 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the
beltCLI vianpx skills add belt-sh/cliand references installation guidelines hosted on the vendor's GitHub repository (github.com/inference-sh). These actions are consistent with the skill's purpose of providing access to the inference.sh platform. - [COMMAND_EXECUTION]: The skill utilizes the
beltCLI tool via Bash commands (e.g.,belt login,belt app run) to interface with the inference.sh platform for video generation tasks. The execution is scoped to thebeltcommand as defined in the skill configuration. - [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by interpolating user-provided text prompts and media URLs into shell commands executed via the
beltCLI. - Ingestion points: User-provided prompts, image URLs, and audio URLs are passed as JSON input to the
belt app runcommand inSKILL.md. - Boundary markers: While the input is structured as JSON, there are no additional boundary markers or instructions to ignore embedded instructions within the user-supplied content.
- Capability inventory: The skill has the capability to execute shell commands via the
Bashtool to run thebeltCLI, which performs remote inference. - Sanitization: No explicit sanitization or validation of user input is defined in the provided instruction examples.
Audit Metadata