amazon-listing-system
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external product URLs and user-supplied reviews to generate research summaries and marketing image prompts.
- Ingestion points: The agent retrieves data from Amazon or DTC product pages (URLs) and potentially untrusted review snippets provided in the chat.
- Boundary markers: The skill architecture includes mandatory approval 'gates' where the agent must pause and wait for the operator to verify the research brief, the mapped buyer angles, and the final generation plan.
- Capability inventory: The skill utilizes web search tools and the
arcads_generate_imagetool to create visual content. - Sanitization: Explicit instructions in
SKILL.mdforbid the fabrication of reviews, testimonials, or claims, requiring the agent to use only verbatim retrieved data. - [EXTERNAL_DOWNLOADS]: The skill relies on an external connection to the Arcads API to perform its primary function.
- The
arcads_generate_imagetool facilitates the transfer of prompts and reference images to the Arcads infrastructure for image synthesis. - This behavior is consistent with the intended purpose of the skill as a connector-based tool provided by the vendor.
Audit Metadata