src/vision_engine.py

def build_vision_note(uploaded_image) -> str | None:
    if uploaded_image is None:
        return None
    return "用户提供了一张图片。后续由 Modal VLM 提取图像内容，并转成角色视角下的观察。"