stats-reporting-audit

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests outsider-provided manuscript content (e.g., reviewer comments, Results/Methods text, figure legends/source-data notes, and tables) that users supply as the “input” to classify/extract/map claims, which means arbitrary free text from the provider user can be processed by the LLM without selecting a trusted specific item.