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
Risk Level
MEDIUM
Analyzed
Feb 13, 2026, 07:45 AM