You can use this webcomponent to simulate LLM agent streaming response to client or hook your own logic to really receive chunked data and render it for the client. Supports multi-chat mode and custom parser logic. Includes a rudimentary diffing to make sure only the last node is re-renderend.