awesome-agentic-ai-zh

awesome-agentic-ai-zh is a structured learning map for people who want to learn AI and AI agents, written mainly in Traditional Chinese with Simplified Chinese and English versions. It gathers scattered projects, teaching material, hands-on exercises and required reading into seven stages, and each stage says what to learn, which exercises to do, which projects to look at and what to check before moving on. The goal is to take you from LLM user to agent-system builder who can read what a framework does, design multi-agent collaboration and write your own MCP server.

After shared foundations (Stage 0 basics such as Python, CLI, git, API and JSON; Stage 1 LLM basics including tokens, APIs and running models locally; Stage 2 prompt design with system prompts, few-shot and chain of thought), the path splits into Track A for CLI power users (comparing six mainstream CLI agents, workflow patterns) and Track B for agent builders (frameworks, writing a ReAct agent, multi-agent design). The Claude Code ecosystem stage (MCP, skills, plugins, SDK) serves both. The README says it curates 145+ projects with ratings, covers local LLM options such as Ollama, llama.cpp, LocalAI and MLX, and estimates a main route of at least 14 to 19 weeks or realistically five to six months at 5 to 8 hours a week. It is MIT licensed.