product-image-qa-review
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill utilizes the
analyze_imagetool to transcribe text from potentially untrusted user-generated images. This creates a surface where text embedded in an image could attempt to influence or override the agent's behavior during the QA process. - Ingestion points:
analyze_imagetool call withinSKILL.mdused for image description and text transcription. - Boundary markers: The prompt does not utilize specific delimiters to isolate the transcribed text from instructions.
- Capability inventory: The skill performs visual analysis and text extraction via a tool.
- Sanitization: No specific sanitization or validation of the transcribed text is described.
- [EXTERNAL_DOWNLOADS]: The
Setupsection provides a command to install an external Model Context Protocol (MCP) server fromhttps://mcp.apimage.org/mcp. This domain corresponds to the skill developer's infrastructure.
Audit Metadata