configuring-ai-agents

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

• Third-party content exposure detected (medium risk: 0.30). The AI agent runtime ingests in-flight record fields mapped into the model’s text/record inputs (and inline files text), which are attacker-controlled when an outsider can submit records to the monitored ingestion source/queue/feed; this required workflow path therefore exposes the agent to outsider-authored free text.