#703 · Primary category: AI Agents & Automation

AgentChat

agent agent-memory agent-skills human-in-the-loop langchain llm mcp mcp-client mcp-server mysql redis

AgentChat 是一个基于 LLM 的智能体交流平台,内置默认 Agent 并支持用户自定义 Agent。通过多轮对话和任务协作,Agent 可以理解并协助完成复杂任务。项目集成 LangChain、Function Call、MCP 协议、RAG、Memory、HITL、Skill、Milvus 和 ElasticSearch 等技术,实现高效的知识检索与工具调用,使用 FastAPI 构建高性能后端服务。

Project last updated:08/27/26

GitHub Stars

875

Forks

102

Contributors

11

License

MIT

Why we included this project

AgentChat is a self-hosted agent workspace rather than a framework you have to wire together yourself. The FastAPI backend and Vue frontend arrive as a complete application, so a team gets a working chat UI, user accounts, and a data dashboard without assembling pieces from scratch. Multi-agent collaboration, RAG-backed knowledge Q&A, and runtime-loaded MCP servers cover the usual production needs, and the three-layer memory keeps long conversations coherent while still remembering user preferences over time. One detail worth calling out is the human-in-the-loop flow that turns an OpenAPI spec into a conversational MCP server, with a person confirming each key decision so automation stays under control. Docker deployment and real documentation make this a practical starting point for small teams that want an actual agent workspace instead of a bare library.

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