Fullstack LangGraph Deep search Quickstart

This project demonstrates a fullstack application using a React frontend and a LangGraph Fastapi powered backend agent.

And you will find a lot of similarities with google deep search and jd agent, because they are both give this project spirit.

Development Notices

1. Don't use BaseMessages or other langchain messages class in return state, because they will trigger unnecessary messages event to the frontend. BTW, they also include a lot of unnecessary meta data, the best practice is to use a simple dictionary to store the data, and it compatible with the openai api used format

2. When use Command or conditional_edge:

   • Command use to goto a node and need to update state at the same time
   • conditional_edge use the state to decide which node to goto, and it doesn't need to update state at the same time.

TODO:

Backend

1. 增加秘塔搜索 API 接口
2. 增加 serper 搜索接口

待定：使用https://r.jina.ai/读取网页内容进行详细了解