observe-ai-production

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s required runtime workflow is based on executing one of the listed Phoenix/trace/prompt-learning/eval-pipeline steps, which ingest production traces and exported failure spans (i.e., free-form text originating from the operating user’s production system and/or other users’ interactions), creating an indirect prompt-injection risk via that trace text fed into the agent/LLM context.