Hello-Agents is an open-source, community-driven tutorial from Datawhale that teaches learners how to build AI agents from the ground up. Rather than treating agent frameworks as black boxes, it walks through the underlying principles—LLM fundamentals, classic reasoning patterns like ReAct and Plan-and-Solve, memory systems, context engineering, and agent communication protocols—so learners understand what frameworks are actually doing under the hood.
The material is organized into 16 chapters across five parts, progressing from agent and LLM foundations, through building LLM-powered agents, into advanced topics such as multi-agent training and evaluation, and finally into comprehensive hands-on projects like an intelligent travel assistant, a research agent, and a simulated "cyber town" of interacting agents. A set of supplementary chapters covers interview preparation, GUI agents, and web agents.
The tutorial is aimed at software engineers, students, and self-learners who already have basic Python skills and some familiarity with large language models, and who want to move from using agent frameworks to understanding and building them. It includes a self-built reference framework (HelloAgents) alongside the code examples, making it practical for readers who learn best by running and modifying real projects.
Access
The full tutorial, source code, and project materials are freely available on GitHub under a Creative Commons BY-NC-SA license, with active community contribution and discussion around the content.