cloudbase-agent

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow exposes agent input text provided by the HTTP caller (AG-UI/OpenAI request body) to the backend LLM via the server adapters (e.g., create_send_message_adapter / create_openai_adapter invoked from AgentServiceApp), so outsider-authored free text is ingested without selecting a specific trusted item first.