luma

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow calls Luma’s public API endpoints (e.g., GET /v1/calendar/list-events, GET /v1/event/get-guests) and ingests the returned JSON fields (event/guest/user names, emails, descriptions) into the agent’s LLM context, which is outsider-authored free text originating from a third-party service.