MeiGen Visual Creative Expert
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill facilitates 'Reference Image' generation by allowing the use of local file paths in the
referenceImagesparameter. While intended for legitimate image-to-image workflows, this capability involves the agent accessing the local filesystem to process files for the generation tool. - [PROMPT_INJECTION]: The skill operates by taking user prompts and gallery search results and interpolating them into tasks for sub-agents (
meigen:image-generator). The instructions do not define specific boundary markers (e.g., XML tags or delimiters) to separate system instructions from untrusted data, which is a common surface for indirect prompt injection, though mitigated by the specialized nature of the sub-agents.
Audit Metadata