stable-diffusion-image-generation

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches pre-trained machine learning models, embeddings, and style adapters from Hugging Face's official model hub.
  • [EXTERNAL_DOWNLOADS]: Provides instructions to install performance-optimizing libraries such as xformers from the official Facebook Research GitHub repository.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied text prompts for image generation in SKILL.md and references/advanced-usage.md.
  • Ingestion points: prompt and negative_prompt parameters in generation functions and API endpoints.
  • Boundary markers: Standard natural language prompt interpolation without specific delimiters.
  • Capability inventory: Local file access for reading/writing images and network access for downloading model weights.
  • Sanitization: The documentation explicitly references the use and configuration of the library's built-in safety_checker to filter outputs.
  • [DYNAMIC_EXECUTION]: Utilizes torch.compile for runtime model optimization to improve inference speed, which is a standard feature of the PyTorch framework.
  • [COMMAND_EXECUTION]: Includes standard instructions for setting up environments via pip and deploying services using Docker and Kubernetes, matching the intended developer-focused use case.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 07:07 PM
Security Audit — agent-trust-hub — stable-diffusion-image-generation