ai-engineering

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

• Third-party content exposure detected (high risk: 0.90). The skill's RAG/multimodal/agents guidance explicitly instructs ingesting and passing retrieved third‑party content into prompts and agent decisions (e.g., "retrieved_chunks", native PDF ingest in rules/multimodal.md, top‑K chunk retrieval in rules/rag.md, and the explicit fetch_url(any url) dangerous-pair mention plus the requirement to classify "every retrieved chunk" in rules/safety-and-guardrails.md), so untrusted external pages/uploads are read/interpreted and can materially influence tool use and next actions.