ma-scout

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to fetch and parse open/public third‑party content (PubMed E-utilities calls, Google Scholar/WebSearch and profile WebFetch, mcp__claude_ai_Consensus and Scholar_Gateway semantic searches, bioRxiv/medRxiv preprint searches, and PROSPERO WebSearch/WebFetch) and then uses those external results to drive gap analysis, ranking, PROSPERO checks, and next actions, so untrusted third‑party content can materially influence agent decisions.