taxonomy-mapping

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

• Third-party content exposure detected (medium risk: 0.65). SKILL.md only describes spawning a subagent and the prompt.md specifies it will use paper_collection fields like titles/abstracts/keywords, which in the required workflow implies runtime ingestion of paper text/metadata provided by an external dataset/collection (outsider-authored content via the subagent input context), creating an indirect prompt-injection surface.