setup-arxiv-mcp
Warn
Audited by Snyk on Feb 13, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). The skill directs the agent to use the "paper-search" MCP to search, download, and read content from public academic sites (arXiv, PubMed, bioRxiv/medRxiv, Semantic Scholar, Google Scholar, CrossRef, etc.), meaning the agent will ingest untrusted third‑party web content that could carry indirect prompt injections.
Audit Metadata