docling

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly supports converting and ingesting documents from arbitrary HTTP/HTTPS URLs (e.g., convert("https://arxiv.org/pdf/2408.09869") in SKILL.md and the "URL Conversion" section in references/parsing.md), so the agent will fetch and parse untrusted public content which can be used in downstream chunking/RAG workflows and thus could enable indirect prompt injection.