m365-agents-py

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow ingests context.activity.text from the inbound chat/message activity (i.e., free-form user/third-party authored message content) into the agent’s LLM-relevant context via handlers like @AGENT_APP.message("message") / send_activity(f"You said: {context.activity.text}"), which can then be forwarded to the model depending on the SDK’s response/streaming path.