#122 · Primary category: Knowledge Base & RAG

HuixiangDou

application assistant assistant-chat-bots chatbot dsl group-chat image-retrieval lark llm multimodal pipeline rag robot wechat

HuixiangDou: Overcoming Group Chat Scenarios with LLM-based Technical Assistance

Project last updated:11/24/25

GitHub Stars

2.5K

Forks

179

Contributors

17

License

BSD-3-Clause

Why we included this project

Support teams and maintainers who field the same questions repeatedly in WeChat, Lark, or other group chats will find HuixiangDou useful. It builds an LLM-backed assistant from your existing documents like PDFs, markdown, code, and spreadsheets. The assistant joins the group and only replies when it is confident it has a good answer, using a three-stage pipeline of preprocessing, rejection, and response so it does not flood the channel. It requires no fine-tuning, runs on CPU-only machines or modest 2 GB GPU setups, and includes complete web, Android, and pipeline source code for industrial deployment. A separate streaming chat mode lets people talk directly to a repository, which helps newcomers learn a large codebase. The project also points to HuixiangDou2, a GraphRAG variant used in plant science research that contributed to a cover paper in Molecular Plant. This is a practical option for teams that want a knowledge assistant wired into the messaging tools they already use.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category