A theory-and-practice tutorial for building AI Native Agents from scratch, covering classic paradigms, multi-agent applications, and agentic reinforcement learning. The primary content is authored in Chinese and covers integrations with several low-code platforms and frameworks.
Project overview
Addresses a gap in systematic, practice-oriented guides for building AI Native Agents by providing instruction from core principles through to multi-agent project development.
Project type
AI Agent · RAG · Prompt Engineering · Learning Resources
Use cases
Knowledge Q&A · Automation · Learning & Education
Deployment
Refer to project documentation
License
License pending
Best for
Developers who read Chinese and want structured, practice-oriented instruction on building AI Native Agents from scratch.
Educators and researchers exploring agent paradigm implementation, multi-agent applications, and Agentic RL training.
Key capabilities
Hands-on implementation of classic agent paradigms including ReAct, Plan-and-Solve, and Reflection.
Develop real-world comprehensive multi-agent projects such as a smart travel assistant, automated deep research agent, and cyber town.
Step-by-step implementation of systematic technologies like context engineering, Memory, protocols, and evaluation.
Master Agentic RL with full-process practical training of LLM from SFT to GRPO.
Limitations and risks
The primary language of the tutorial content is Chinese.
GPU requirements, data boundary specifics, telemetry policies, and external service dependencies are not documented.
Getting started
No installation is required to start reading. Users can begin by visiting the online reading website or downloading the PDF directly.
Alternatives and comparisons
Provides Chinese-translated and reproduced LLM tutorials covering prompt engineering, RAG, and fine-tuning, optimized for Chinese language contexts.
Provides comprehensive programming learning roadmaps, practical project tutorials, and career guidance for beginners learning programming and AI development.
Provides a PyTorch-like library to build and auto-optimize LM workflows via a unified auto-differentiative framework for prompt optimization.