email-design
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references CLI setup instructions hosted on the official GitHub repository of the inference-sh organization.
- [COMMAND_EXECUTION]: The skill uses the belt CLI tool to perform authentication and execute specific image-generation applications.
- [REMOTE_CODE_EXECUTION]: The belt app run command initiates the execution of remote containerized applications (e.g., html-to-image, flux-dev-lora) on the inference.sh platform.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text or design requirements to generate code for visual assets. (1) Ingestion Points: User-defined email content and design prompts. (2) Boundary markers: Not utilized. (3) Capability inventory: Bash execution with the belt tool. (4) Sanitization: Not explicitly performed. The risk is assessed as minimal because the resulting output consists of static images.
Audit Metadata