idea-requirement

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

• Third-party content exposure detected (high risk: 0.85). This workflow explicitly allows each node to perform Web Search at runtime (“Allow each node to call Web Search according to its research policy”), and the research policy requires extracting facts from fetched public pages into stage documents and updating sources/sources.json, which then gets passed into subsequent nodes’ LLM context—thereby ingesting outsider-authored free text from public web content.