Resources
Resources

Hello-Agents: Datawhale's "Building Agents from Scratch" Systematic Tutorial

GitHub hot (Trendshift-listed) Datawhale free Chinese tutorial. AI Native Agent principles (ReAct/Plan-Solve/Reflection code implementation) + self-built framework HelloAgents + context engineering/Memory/Agentic RL + real projects. From "using tools" to "building tools."

Published July 25, 20263 min read
<!-- hello-agents-resource | resource | Hello-Agents Agent Tutorial -->

What Is Hello-Agents: Datawhale's Systematic Agent Learning Tutorial

Hello-Agents is Datawhale community's "Building Agents from Scratch" tutorial, from foundational theory to practice, comprehensively mastering AI Native Agent system design and implementation. Trendshift-listed, free in Chinese.

Core Content

  • AI Native Agent: focuses on truly AI-driven agents (not Dify / Coze flow-driven), piercing through framework surfaces, starting from core principles.
  • Classic paradigm implementation: ReAct, Plan-and-Solve, Reflection-hand-coded implementation.
  • Self-built framework HelloAgents: builds an agent framework from scratch on OpenAI native API.
  • Advanced skills: context engineering, Memory, protocols, evaluation.
  • Agentic RL: full-pipeline hands-on LLM training from SFT to GRPO.
  • Real projects: smart travel assistant, cyber town, and other comprehensive cases.
  • Job interviews: agent-related interview questions.

How to Use

Who It's For

  • AI learners: want to evolve from LLM "users" to Agent "builders," systematically learning agent principles + practice.
  • Job seekers: agent interview questions + real projects, a job-hunt boost.
  • Teaching: Datawhale community tutorial, suited for courses / self-study.

vs. Other Agent Tutorials

Most tutorials stop at Dify / Coze low-code usage; Hello-Agents dives into AI Native Agent principles (ReAct / Plan-Solve / Reflection code implementation + self-built framework + Agentic RL), from "using tools" to "building tools."

Repo: https://github.com/datawhalechina/hello-agents

Related