The gist

Hello-Agents is an open-source tutorial from the Datawhale community that provides a comprehensive guide to building AI-native agents. It addresses the scarcity of systematic, hands-on learning resources in the field. The project takes users from core principles and classic paradigms through to developing their own multi-agent applications, aiming to turn LLM users into agent system builders.

What it does

Teaches core agent concepts, history, and classic paradigms like ReAct and Plan-and-Solve.
Guides you through building a custom agent framework from scratch using native OpenAI APIs.
Provides hands-on experience with low-code platforms like Coze and Dify.
Walks through the application of mainstream frameworks like AutoGen and LangGraph.
Includes advanced topics like memory, RAG, agent communication protocols, and performance evaluation.
Culminates in developing real-world projects like an intelligent travel assistant.

How it works

Hello-Agents is a free, open-source tutorial delivered as a GitHub repository with structured text chapters and corresponding Python code. Users can read the guide online, download a PDF, or clone the repository to run the code locally. The project guides developers through theoretical concepts and hands-on exercises, culminating in the creation of complete multi-agent applications. It is completely free under a Creative Commons license.

Best for

This project is best for developers, students, and researchers with basic Python skills who want a structured, in-depth course on building AI-native agents, from core concepts to implementing complex multi-agent systems.

Watch out for

The documentation and community discussions are primarily in Chinese, though some English translations are available. This is an educational course, not a pre-built, production-ready framework.