feature-store

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

• Third-party content exposure detected (medium risk: 0.30). The workflow ingests free text submitted by outsiders into runtime input fields (e.g., FastAPI /score POST body ScoreRequest with order_values used as order_value for the @on_demand_feature_view computation), so the agent’s required retrieval/feature computation path can consume attacker-controlled text.