paper-writing

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

• Third-party content exposure detected (high risk: 0.80). The skill's Integration with MCP Tools section explicitly instructs the agent to "Search papers via academic-mcp (19 databases) or semantic-scholar-mcp" and to "use arxiv-latex-mcp to read LaTeX source," which requires fetching and reading public third‑party content (arXiv/Semantic Scholar) that the agent will interpret and use in its workflow, enabling indirect prompt injection.