agent-craft
Agent Craft is a systematic open-source educational project designed to guide developers from zero to building full-stack AI agents using Python. It provides runnable code examples and comprehensive blog tutorials covering the entire development lifecycle, from basic LLM prompt engineering to advanced deployment. The curriculum includes 15 modules structured into foundational, framework, and intelligence sections. Key technologies taught include LangChain for building chains and agents, Retrieval Augmented Generation (RAG) for private knowledge base integration, and LangGraph for creating stateful, debuggable multi-agent workflows with human-in-the-loop capabilities. The course also covers Model Context Protocol (MCP) for standardizing agent-tool communication. Each module is self-contained with clear comments and local execution support, ensuring learners can independently learn and reproduce the code. Accompanying articles explain the underlying design principles and mechanisms, helping users move beyond si