paper-to-intuition

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly asks to "Get the paper
• Ask for the arXiv link, PDF, or paper title," meaning the agent will fetch and ingest public third-party content (arXiv or arbitrary PDFs/URLs) that is untrusted/user-provided and will be read/interpreted as part of its workflow, exposing it to indirect prompt injection.