claude-agent-course

type: course-index tags: [AI, Agent, LLM, Course, Anthropic] created: 2026-04-23 --- # Tool List > A comprehensive learning path from zero foundation to independently designing an Agent system. > Approximately **60 lessons**, each taking 10-15 minutes to read, with English instruction + life-like analogies + an "Aha moment" in each lesson. > 🌍 **Open Source**: This tutorial is open-sourced under the **MIT License** at [github.com/recomby-ai/claude-agent-course](https://github.com/recomby-ai/claude-agent-course). Native Obsidian format (WikiLinks `[[xxx]]`), recommended to clone into Obsidian vault for reading; GitHub webpage can also be used to view content directly. ## What This Tutorial Solves You've browsed Anthropic blogs, YouTube, Twitter, and various public accounts, but the information is scattered. This tutorial does three things: 1. **Starts from scratch** — No prior knowledge of tokens, context windows, or embeddings is assumed; foundational concepts are thoroughly covered. 2. **Organizes Anthropic's official content and industry blogs into a structured path** — Not just a list of articles, but a course. 3. **Each lesson has an Aha moment** — Humorous, analogical, and counterintuitive insights, not just academic translations. ## Learning Path (3 Tracks) ``` 🚀 Track 1 · Getting Started (~3 days to read through) Module 0 → 1 → 2 You'll learn: LLM / API / Basic Agent Form ⚙️ Track 2 · Advanced (~1 week) Module 3 → 4 → 5 → 6 You'll learn: Agent Architecture, Harness, Claude Code, MCP 🏭 Track 3 · Practical (in-depth as needed) Module 7 → 8 → 9 → 10 You'll learn: Eval, RAG, Multi-Agent Frameworks, Production Environment 📚 Track 4 · Supplemental (optional) Module 11 → 12 Core papers + industry classics ``` ## Complete Module List | # | Module | Lessons | Topic | | --- | ------------------------------------------- | ------- | ------------------------------------------------------------ | | 0 | [[Module 0 - 基础积木/_overview\|Module 0 Overview]] | 6 | LLM API / Token / Embedding / Agent Evolution History | | 1 | [[Module 1 - Tool Use/_overview\|Module 1 Overview]] | 6 | Giving LLMs Hands and Eyes (reconstructed from "Five-Ring Tutorial") | | 2 | [[Module 2 - Context Engineering/_overview\|Module 2 Overview]] | 5 | Don't Give AI a 1000-Page Manual | | 3 | [[Module 3 - Agent 架构/_overview\|Module 3 Overview]] | 5 | Workflow vs Agent / 5 Modes / Multi-Agent | | 4 | [[Module 4 - Harness/_overview\|Module 4 Overview]] | 4 | Preventing Agent Failure at Step 50 | | 5 | [[Module 5 - Claude Code 和 Agent SDK/_overview\|Module 5 Overview]] | 4 | Practical Tools | | 6 | [[Module 6 - MCP/_overview\|Module 6 Overview]] | 3 | The USB Interface of Tools | | 7 | [[Module 7 - Eval/_overview\|Module 7 Overview]] | 4 | How to Know if Your Agent Really Works | | 8 | [[Module 8 - RAG 体系/_overview\|Module 8 Overview]] | 4 | From Retrieval Enhancement to Retrieval Agents | | 9 | [[Module 9 - Multi-Agent 框架/_overview\|Module 9 Overview]] | 3 | LangGraph / CrewAI / AutoGen / Swarm | | 10 | [[Module 10 - 生产环境/_overview\|Module 10 Overview]] | 6 | Token Economy / Caching / Observability / Security / Deployment | | 11 | [[Module 11 - 核心论文/_overview\|Module 11 Overview]] | 5 | ReAct / Reflexion / Toolformer / ToT / Agent Survey | | 12 | [[Module 12 - 社区经典/_overview\|Module 12 Overview]] | 5 | Simon Willison / Karpathy / Lilian Weng / Hamel / Eugene Yan | | 13 | [[Module 13 - 附录/_overview\|Module 13 Overview]] | 4 | Glossary / Original List / Cookbook Index / Courses Index | ## Fixed Structure for Each Lesson ``` # [N.M] · [Title] ## See a Scenario ← Life Analogy ## Core Concept in One Sentence ← Clearly Explained ## Walkthrough ← Code/JSON/Example ## ⚠️ Common Pitfalls ## 💡 Aha Moment ← Counterintuitive Insights ## Self-Test ## Next Step ← Link to Next Lesson ``` ## How to Use This Tutorial - **Don't Read from Start to End**. Follow the Track diagram and jump to sections of interest. - **Each Lesson is an Independent Unit**. You can stop after completing a lesson. - **Self-Tests are Not Optional**. If you fail, revisit the material. - **2-3 Lessons per Day** is better than **10 Lessons in One Day**. ## Upstream: Three-Level Evolution Chain All content in this tutorial can be mapped to [[../../Prompt Engineering|Prompt Engineering]] → [[../../Context Engineering|Context Engineering]] → [[../../Harness Engineering|Harness Engineering]] , a three-level evolution chain. The former three are **concept maps** (bird's-eye view), while this tutorial is **ground-level marching** (step-by-step walking). ## Related - [[../README|↑ AI Agent Development Area]] - [[../Prompt Engineering]] — Concept Map Level 1 - [[../Context Engineering]] — Concept Map Level 2 - [[../Harness Engineering]] — Concept Map Level 3