planning

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

• Third-party content exposure detected (medium risk: 0.30). In planning Step 1 the agent “read[s] relevant design docs, RFCs, ADRs, READMEs, referenced library/API docs, and any spec the user points to,” and those user-pointed documents (and user request text) are free-form outsider-authored input that the runtime LLM ingests before narrowing to a specific trusted item.