arxiv-monitor
Warn
Audited by Snyk on Apr 18, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.70). The skill's scripts (scripts/arxiv_monitor.py -> _run_search calling the sibling arxiv-search script/SEARCH_SCRIPT) and the SKILL.md explicitly query the public arXiv API and ingest returned paper metadata/abstracts (public, user-submitted content) which the agent reads and uses to decide which papers to present or analyze, so untrusted third‑party content can influence actions.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata