deepread

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

• Third-party content exposure detected (high risk: 0.90). The skill ingests arbitrary user-provided PDFs and returns/exposes their extracted text and structured data in its runtime workflow (see the response examples with "result.text"/"result.data", the webhook payloads, and the Public Preview URLs like https://preview.deepread.tech/Xy9aB12), so untrusted third‑party content can be fed into and influence the agent's decisions.