Hello-Agents is a comprehensive, open-source tutorial project from the Datawhale community designed to teach users how to build intelligent agents from first principles. The repository is written in Python and serves as a systematic learning guide that bridges the gap between agent theory and practical implementation. The project addresses a critical gap in the market, as the README notes that while 2024 was the year of "hundred models competing," 2025 marks the beginning of the "Agent era," yet systematic, practice-focused tutorials remain scarce.
The project distinguishes between two primary approaches to agent construction: software engineering-driven agents like Dify, Coze, and n8n, which are process-driven with LLMs serving as data processing backends, and AI-native agents that are truly driven by artificial intelligence. Hello-Agents focuses on teaching the latter approach, aiming to help learners understand core principles, penetrate framework abstractions, master classical paradigms, and ultimately build their own multi-agent applications. The tutorial emphasizes hands-on learning as the most effective educational method.
The curriculum is structured across five major parts spanning sixteen chapters. The first part covers foundational concepts including agent definitions, types, paradigms, the historical evolution from symbolic AI to LLM-driven agents, and large language model fundamentals. The second part guides learners through building LLM-based agents, including classical pattern construction like ReAct and Plan-and-Solve, low-code platform usage such as Coze and Dify, mainstream frameworks like AutoGen and LangGraph, and building custom agent frameworks from scratch. The third part extends into advanced topics including memory systems, retrieval-augmented generation, context engineering, agent communication protocols like MCP, agentic reinforcement learning from SFT to GRPO, and agent performance evaluation. The fourth part presents comprehensive case studies including an intelligent travel assistant, automated deep research agents, and a cybernetic town simulation. The final section covers graduation projects and future perspectives.
According to GitGenius tracking data, the repository has demonstrated steady growth with stargazers increasing from 63,930 to 63,931 between checks. The project maintains exceptionally responsive issue and pull request handling with a median latency of zero hours and a mean of 1.2 hours across 286 tracked items. Documentation represents the most active issue label with 268 entries. The primary contributor jjyaoao has logged 481 events, with additional contributors hhh-king and Jack-ctrl6 contributing 12 and 10 events respectively. The project overlaps with other significant repositories including PaddlePaddle's PaddleOCR, LlamaFactory, and Dify through shared contributors.
The repository offers learners free access to all content, hands-on implementation experience, a custom HelloAgents framework built from OpenAI's native API, advanced techniques including context engineering and memory systems, practical training in agentic reinforcement learning, real-world case studies, and interview preparation materials. The project includes an Extra-Chapter section with community contributions covering topics like Dify tutorials, GUI agents, web agents, and practical development experiences. The online reading platform is available both internationally and with domestic acceleration for Chinese users, eliminating the need for local downloads.
