deepsearch-agents
🔎 深度研搜对话式多智能体 AI Agents,最适合系统学习 DeepAgents 的实战项目|AI Deep Research Agent 实战 · LangGraph + RAGFlow + Tavily + FastAPI + WebSocket 从0
What is deepsearch-agents?
🔎 深度研搜对话式多智能体 AI Agents,最适合系统学习 DeepAgents 的实战项目|AI Deep Research Agent 实战 · LangGraph + RAGFlow + Tavily + FastAPI + WebSocket 从0到工程化落地。前后端完整代码全栈可跑,Docker 环境一键部署,配套 ai-agents-from-zero 免费教程与章节代码分支。适合系统学习大模型应用、多智能体 Agent 和企业级 AI 工程落地
Overview of deepsearch-agents
What it is
deepsearch-agents is a conversational multi-agent research system designed for deep research scenarios. It uses DeepAgents to organize a main agent and expert sub-agents, enabling the system to search public networks, query structured databases, access private knowledge bases via RAGFlow, read user-uploaded attachments, and deliver results as answers, Markdown, or PDF. The project is the companion source code for the ai-agents-from-zero tutorial series, providing a complete runnable full-stack implementation with Docker one-click deployment.
Key features
- Orchestrator-Workers architecture: one main agent for planning and dispatching, three expert sub-agents for internet search, database query, and RAGFlow knowledge base.
- Multi-source retrieval: Tavily for public web, MySQL for structured data, RAGFlow for internal documents, and file reading for uploaded attachments.
- Full deliverable pipeline: from retrieval to Markdown/PDF generation, not just prompt design.
- Observable long-running tasks: tool calls, sub-agent calls, results, cancellations, and errors are pushed to the frontend via WebSocket.
- Session-level context isolation: thread_id and session_dir distinguish tasks; ContextVar provides identity and file directory to deep tools.
- Engineering-grade frontend and backend: FastAPI + WebSocket + DeepAgents + React with clear separation.
- Free systematic tutorial: chapter-by-chapter Git branches and online documentation accompany the code.
Use cases
- Learning DeepAgents systematically beyond toy examples.
- Understanding how to integrate Tavily, MySQL, RAGFlow, and LLMs in a single research assistant scenario.
- Building an AI Agent project that is closer to real-world development than simple model calls.
- Creating a portfolio project with clear layers: agent, tool, service, file, and frontend.
Getting started
See the repository README for installation and quick-start commands.
License and pricing
The license is not stated in the repository metadata. The project is listed as open source.
Based on the project's public README and repository metadata.