image-generation-enhanced
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text descriptions to generate image prompts, creating a potential surface for indirect prompt injection where instructions could be embedded in user input.\n
- Ingestion points: User requests for image generation and reference image descriptions.\n
- Boundary markers: The skill encourages the use of structured JSON blocks to encapsulate prompt parameters, which can help delimit user-supplied intent from the overall prompt structure.\n
- Capability inventory: Uses external API calls (OpenRouter) and local CLI tool execution (minibanana).\n
- Sanitization: No specific sanitization or filtering logic for user input is described in the guidelines.
Audit Metadata