wtf is agent-study?

This is a 36-chapter course on building AI agents, written in Chinese and targeted at developers who want to learn AI agent engineering for job interviews and production use. An AI agent is a program that uses a language model not just to answer questions, but to plan, use tools, remember context across steps, and carry out multi-step tasks autonomously. The course progresses through seven layers of depth, starting with fundamental theory like the ReAct loop (a pattern where the AI reasons then acts, then reasons again based on what happened) and moving through practical frameworks, production infrastructure, advanced architectures, and expert-level topics. Each of the 36 chapters is a standalone runnable Python file that serves as both a lesson and working code you can execute and modify. Topics covered include how to build multi-agent systems where several AI agents collaborate, retrieval-augmented generation (giving agents access to searchable knowledge bases), the MCP protocol for standardizing how agents call external tools, agent memory systems, security against prompt injection attacks, how to monitor and trace agent behavior in production, and how to fine-tune models for tool use. You would use this course if you are a Chinese-speaking developer wanting a structured, code-first introduction to AI agent development that prepares you for technical interviews. Most chapters can run without an AI API key. Requires Python 3.10 or later.