agent-platform-rag-engine-management

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

• Third-party content exposure detected (high risk: 0.70). The required runtime workflow uses Vertex AI RAG Engine retrieval (rag.retrieval_query(...) and/or the RAG tool in generate_content), which ingests the corpus’s stored document text (e.g., file contents from the specified ragCorpora/{corpus_id}) into the LLM context; if that corpus contains outsider-authored documents, their free text can be fed to the model via context.text / tool grounding.