pilot-dataset

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow uses pilotctl --json send-message / send-file to exchange dataset metadata and files with a peer, so the agent will ingest free-form dataset content/metadata originating from an OUTSIDER peer (e.g., the publisher’s dataset schema/rows and file body) into the LLM context for processing/validation.