amz-product-research

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow requires collecting “what the user can observe about the top listings (reviews, ratings, image quality, complaints)” and “any known patents, gating, or compliance issues,” which in practice typically involves ingesting outsider-authored free-form text from third-party sources like scraped/public Amazon listing reviews or other web content into the LLM context.