research-proposal

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

• Third-party content exposure detected (high risk: 0.90). The skill's Phase 2 "Literature Collection" explicitly instructs the agent to perform WebSearch and ingest content from open/public sources (e.g., arXiv, PubMed, bioRxiv, SSRN, Google Scholar) and to retrieve full texts via Zotero MCP, so the agent will read untrusted third‑party web/preprint content as part of its workflow.