hello-agents

Hello-Agents is an open-source intelligent agent system construction tutorial from the Datawhale community, positioned as a zero-based learning guide that balances theory and hands-on practice, focusing on AI-native intelligent agents rather than process-driven software engineering agents. The tutorial is divided into five major parts with a total of 16 chapters: Part 1 covers the definition of intelligent agents, their development history and basic large language model knowledge; Part 2 guides readers to implement classic paradigms including ReAct, Plan-and-Solve and Reflection, introduces low-code platforms such as Coze, Dify and n8n, as well as frameworks AutoGen, AgentScope and LangGraph, and builds a self-developed framework HelloAgents from scratch; Part 3 covers memory and retrieval (RAG), context engineering, MCP/A2A/ANP communication protocols, Agentic-RL (from SFT to GRPO) and performance evaluation; Part 4 features comprehensive practical cases, including an intelligent travel assistant, reproduction of DeepResearch Agent, and cyber town social simulation; Part 5 is dedicated to graduation project design. Selected community contributions include summaries and answers to agent interview questions, supplementary context engineering content, a step-by-step Dify tutorial, comparison between Agent Skills and MCP, hands-on practice of GUI Agent and Web Agent, best practices for skill writing, agent self-evolution, post-training of the travel assistant and other extended content. The project provides online reading, PDF download and supporting code, and adopts the CC BY-NC-SA 4.0 license agreement.