langgraph-orchestration

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

• Third-party content exposure detected (medium risk: 0.30). The runtime ingests outsider free text in state["messages"] (e.g., via HumanMessage(...)) and passes it directly to the LLM in the chatbot, supervisor, and tool-enabled agents (llm.invoke(...) / llm_with_tools.invoke(...)), so an outsider can submit arbitrary prompt text that the workflow consumes.