systematic-review-prisma

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly requires ingesting and summarizing raw search results and sources (see SKILL.md step 3 "Export raw search results to sources/metadata/" and the references/mcp-catalog.md which lists arXiv, Semantic Scholar, OpenAlex, publisher pages and web search/Google Scholar as data sources), so the agent will read untrusted public third‑party content that can influence screening, inclusion decisions, and downstream actions.