argument-mapping

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow is described as extracting claims from “source material” and linking evidence, which implies it may ingest outsider-authored free text from external sources (e.g., public web pages or other documents) into the LLM context via the claim-extraction/evidence-linking steps; no exclusion is specified that limits sources to user-authored content.