j-cover
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-uploaded images as a primary input source, which represents a potential surface for indirect prompt injection if the agent attempts to follow text instructions embedded within the image (OCR injection).
- Ingestion points: User-provided image referenced in the 'INPUT IMAGE' and 'PREREQUISITE' sections of SKILL.md.
- Boundary markers: Absent; the instructions do not explicitly tell the agent to ignore or delimit instructions that might be contained within the source image.
- Capability inventory: The skill's capabilities are restricted to image generation, outpainting, and layout design; it lacks access to the file system, network, or shell command execution tools.
- Sanitization: Absent; the skill does not specify any validation or filtering of text derived from the input media.
- [SAFE]: The skill's instructions use emphatic language (e.g., 'CRITICAL', 'MUST') to reinforce design constraints and maintain character consistency, which is standard instructional behavior and does not constitute a prompt injection or safety bypass attempt.
Audit Metadata