ai-video-generation
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the installation of the
belt-sh/clitool and related components using thenpx skills addcommand. It also provides links to installation scripts hosted on a public GitHub repository (github.com/inference-sh/skills). - [COMMAND_EXECUTION]: The skill utilizes
Bashto executebeltCLI commands. This execution is restricted by the skill's configuration to only allow thebeltbinary, implementing a principle of least privilege. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided prompts that are passed as arguments to video generation models. This creates a surface for indirect prompt injection where crafted inputs could attempt to manipulate the generation output or the agent's execution flow. * Ingestion points: User prompt strings in
belt app runcommands (SKILL.md). * Boundary markers: No specific delimiters or safety warnings are included to isolate user input. * Capability inventory: Access toBashrestricted to thebeltCLI. * Sanitization: The instructions do not define any sanitization or validation logic for external inputs.
Audit Metadata