ml-paper-writing

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 search and ingest open web content (e.g., via Exa MCP at https://mcp.exa.ai/mcp and APIs like Semantic Scholar, CrossRef, and arXiv) as a mandatory part of the citation-verification and literature-search workflows in SKILL.md and references/citation-workflow.md, so untrusted third‑party content will be read and can influence drafting and tool actions.