ai-engineer

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

• Third-party content exposure detected (medium risk: 0.65). The /ai skill’s required runtime workflow is to consult workflow-engine and then run an LLM agent/RAG loop; this workflow commonly ingests outsider-authored free text via retrieved web/document content (RAG chunks) or user-provided messages from parties other than the operating user, which the LLM reads as context.