read-arxiv-paper

Warn

Audited by Snyk on Feb 15, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.80). The skill directly downloads and unpacks TeX source from arbitrary public arXiv URLs (e.g., https://www.arxiv.org/src/{arxiv_id}) and reads/interprets those files to produce summaries, exposing the agent to untrusted third-party content that could embed instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 15, 2026, 08:16 PM