arxiv-batch-reporting

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly states the skill "leverages the arXiv API" to fetch and batch-process arXiv preprints (e.g., query profiles, result retrieval, and abstract-based relevance/novelty scoring), which are public, third‑party user-generated documents the agent reads and uses to drive ranking, alerts, and follow-up actions.