p-image
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and command examples are designed for legitimate use of the inference.sh platform via its official CLI tool.\n- [EXTERNAL_DOWNLOADS]: The documentation references installation scripts and related skills hosted on the official inference-sh GitHub repository.\n- [COMMAND_EXECUTION]: The skill uses the
infshCLI tool to perform model inference, which is limited by the skill's configuration and consistent with its stated purpose.\n- [PROMPT_INJECTION]: The skill processes user-supplied text prompts and image URLs for generative tasks, which is a potential surface for indirect prompt injection.\n - Ingestion points: Text prompts and image URLs passed as arguments to the
infshcommand inSKILL.md.\n - Boundary markers: Input data is structured within JSON objects passed to the command-line tool.\n
- Capability inventory: The skill utilizes the
infshtool (Bash) to perform external API requests for image generation as specified inSKILL.md.\n - Sanitization: Content filtering and sanitization are expected to be performed by the inference service and the underlying AI models.
Audit Metadata