medical-advisory

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

• Third-party content exposure detected (high risk: 0.80). The skill's MCP retrieval workflow in SKILL.md and the evidence-mining.py script show the agent will call external public sources (PubMed, OpenAlex and Metaso which includes "paper/webpage") to fetch and interpret third‑party web/paper content as part of decision-making, so untrusted external content can materially influence actions.