radar

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

• Third-party content exposure detected (high risk: 0.80). The skill instructs the agent to fetch and run actions against the Radar API via Membrane (e.g., "membrane action run" and "membrane request CONNECTION_ID /path/to/endpoint"), and the documented data types include user-generated items like Forum Post and Chat Message, so the agent will ingest third-party/untrusted content that could influence subsequent actions.