llm-models
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references official installation guides and instructions hosted on the inference.sh GitHub repository. These are standard resources for setting up the documented CLI tool.
- [INDIRECT_PROMPT_INJECTION]: The skill accepts user-defined prompts via JSON input which are passed to external LLM providers for inference. This represents a standard input surface for AI interactions.
- Ingestion points: User-supplied
promptfield within the JSON input inSKILL.md. - Boundary markers: Input is structured using JSON, providing clear separation between model parameters and the prompt text.
- Capability inventory: The
belttool is permitted to perform network requests for LLM inference. - Sanitization: Relies on the inference.sh CLI and the target LLM providers' internal safety filters.
Audit Metadata