Course
AI Agent Engineering
A comprehensive course for backend engineers learning AI Agent development: manual agent construction, AI-assisted development, MCP/LSP/A2A protocols, task orchestration, and production engineering.
40 chaptersPython / Node / Go / JavaMCP / LSP / A2A40 chaptersRunnable labs
Stage 1 / 1-2 weeks
Agent 核心与古法手写
Agent Core — Manual Implementation
Chapter 1: 从函数到 Agent：概念与边界From Functions to AgentsChapter 2: 环境配置与第一次 LLM 调用Environment and First LLM CallChapter 3: LLM 推理参数全解LLM Call Parameters ExplainedChapter 4: 手写 ReAct：思考-行动-观察循环Handwritten ReAct LoopChapter 5: Tool Calling 从零实现Tool Calling from ScratchChapter 6: 记忆系统：上下文与向量检索Memory: Context and VectorsChapter 7: 规划与自我纠错Planning and Self-RecoveryChapter 8: Agent 评估：如何判断做对了How to Evaluate an Agent
Stage 2 / 1-2 weeks
协议生态：MCP / LSP / A2A
Protocol Ecosystem
Chapter 9: Function Calling 与 JSON Mode 深度解析Function Calling vs JSON ModeChapter 10: MCP 协议导论MCP Protocol PrimerChapter 11: 手写 MCP Server（stdio）Handwritten MCP Server (stdio)Chapter 12: MCP Client 与 SSE/HTTP TransportMCP Client and TransportsChapter 13: LSP 协议与最小实现LSP Protocol and Minimum ServerChapter 14: LSP + Tree-sitter：让 Agent 读懂代码LSP + Tree-sitter for AgentsChapter 15: A2A 协议：Agent 间协作A2A: Agent-to-Agent ProtocolChapter 16: 协议选型与组合架构When to Use Which Protocol
Stage 3 / 1 week
AI 辅助开发
AI-Assisted Development
Chapter 17: Prompt-First DevelopmentPrompt-First DevelopmentChapter 18: 上下文工程：规则文件与代码库理解Context Engineering RulesChapter 19: AI 辅助编码工具链对比AI Coding ToolchainsChapter 20: 用 AI 生成后端代码AI-Generated Backend CodeChapter 21: AI 辅助代码审查AI-Assisted Code ReviewChapter 22: Vibe Coding 的边界Limits of Vibe CodingChapter 23: AI 生成代码的验证策略Verifying AI-Generated CodeChapter 24: 实战：用 AI 辅助重构后端服务Practice: AI-Assisted Refactor
Stage 4 / 1-2 weeks
任务编排与多 Agent 系统
Task Orchestration and Multi-Agent Systems
Chapter 25: 工作流编排：从状态机到 DAGFrom State Machines to DAGsChapter 26: 多 Agent 协作模式Multi-Agent Collaboration PatternsChapter 27: 长任务：幂等、重试、超时、中断恢复Idempotency, Retry, Timeout, RecoveryChapter 28: 人工介入点Human-in-the-LoopChapter 29: LangGraph 实战LangGraph PracticeChapter 30: Temporal 与事件驱动编排Temporal and Event-Driven FlowsChapter 31: 实战：代码审查 Agent 流水线Practice: Code Review Agent PipelineChapter 32: 实战：自动化运维 AgentPractice: Automated SRE Agent
Stage 5 / 1 week
工程化、部署与安全
Production Engineering, Deployment, and Security
Chapter 33: Agent 安全Agent SecurityChapter 34: 测试策略与 EvalsTesting and EvalsChapter 35: 可观测性ObservabilityChapter 36: 成本优化与缓存Cost Optimization and CachingChapter 37: 部署架构Deployment ArchitecturesChapter 38: 多语言 Agent 实现对比Multi-Language Agent ComparisonChapter 39: 课程大项目：自动化后端开发助手Capstone: Backend Development AssistantChapter 40: 总结与进阶路线Summary and Advanced Roadmap