muapi-rednote-cover
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill incorporates user-provided text from the
topic,style, andtext_overlayinputs directly into the prompts sent to the image generation models. - Ingestion points: User inputs defined in the
Inputssection ofSKILL.mdare used in the prompt construction in Phase A. - Boundary markers: The instructions do not use delimiters or specific safety warnings to isolate user-supplied content from the system instructions.
- Capability inventory: The skill utilizes the
muapiCLI tool and standardcurlcommands to interact with external image generation APIs. - Sanitization: There is no evidence of sanitization or escaping of user-provided strings before they are interpolated into the prompt templates.
Audit Metadata