LangGraph Visualization validation

[tmdqja75]

I am building an application using LangGraph + OpenAI model. I was trying to visualize the graph that I build for debugging purpose using the following graph function

graph.get_graph().draw_mermaid_png()

Here is the graph code

    workflow = StateGraph(AgentState)
    workflow.add_node(
        "agent", partial(agent, model=model)
    )  
    retrieve = ToolNode(toolset)
    workflow.add_node("retrieve", retrieve)

    workflow.add_conditional_edges(START, partial(check_relevance, model=model))
    workflow.add_conditional_edges(
        "agent",
        tools_condition,
        {
            "tools": "retrieve",
            END: END,
        },
    )
    workflow.add_edge("retrieve", END)
    graph = workflow.compile()

and here is the final graph visualization:

I'm confused. Why is there a conditional edge in between START -> retrieve node? The only conditional edges I used in my graph are in between START -> agent, END (with check_relevance conditional function), and agent -> retrieve, END. Is this a bug on visualization side?