image-generation

Pass

Audited by Gen Agent Trust Hub on Oct 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Command Execution: The skill invokes several local tools and scripts to facilitate its workflow, such as generate_image for asset creation and a Node.js script located at /workspace/tools/review-thumb.mjs for generating review sheets. It also mentions utility tools like hyperframes and image-ops. These operations are used as intended within the local workspace environment to process and manage generated assets.
  • Indirect Prompt Injection Surface: The skill processes user-provided descriptions to construct prompts for an image generation model. This interaction represents a surface where external input influences the behavior of a tool. The skill includes instructions to mitigate risks by following provider-level moderation messages and adhering to specific structural constraints defined in the documentation.
  • Workspace Data Access: To ensure brand consistency, the skill is designed to read configuration files and design tokens from the /workspace/brand/ directory. This access allows the agent to incorporate project-specific visual requirements, such as hex color palettes and tonal guidelines, into its generation process.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 28, 2026, 05:18 AM