paper2blog

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

• Third-party content exposure detected (high risk: 0.85). High likelihood: the runtime workflow ingests outsider-authored free text from the input paper PDF via scripts/extract_pdf.py into assets/meta/text.txt and assets/meta/captions.json, which the agent then uses to draft the articles (prompt context includes that extracted prose).