Summary

Agent Craft is a systematic, open-source educational project designed to guide developers through building full-stack AI agents. It targets individuals who understand AI agent concepts but need practical, hands-on experience, offering a clear progression from basic LLM calls to advanced RAG, LangGraph, and deployment. The project's benefit lies in its runnable code examples and detailed blog explanations, demystifying the underlying principles of agent decision-making and action.

How It Works

The project follows a progressive, modular curriculum, starting with foundational LLM interactions and function calling. It then integrates core AI agent frameworks, including LangChain for tool use and memory management, Retrieval-Augmented Generation (RAG) for incorporating external knowledge, and LangGraph for creating stateful, debuggable agent workflows. This approach emphasizes both practical implementation with runnable, annotated code and theoretical understanding through accompanying blog posts, allowing users to grasp how agents process information, make decisions, and execute tasks.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/Annyfee/agent-craft.git), navigate into the directory, install dependencies (pip install -r requirements.txt), and then install the project in developer mode (pip install -e .).
• Prerequisites: Requires Python 3.10–3.12 and Node.js v20+. API keys for services like OpenAI, LangSmith, and potentially others (e.g., Gaode Maps, ChatGPT) must be configured in a .env file.
• Running: Example scripts can be executed directly, such as python "m01_agent_introduction/Agent-demo.py".
• Links: GitHub repository: https://github.com/Annyfee/agent-craft.git.

Highlighted Details

• Features a structured 15-module learning path covering agent fundamentals, framework integration (LangChain, LangGraph), RAG, multi-agent communication (MCP), and deployment.
• Provides practical, runnable code examples for each module, designed for easy local replication.
• Integrates multiple key technologies: LLMs, Function Calling, LangChain, RAG, LangGraph, MCP, Streamlit for UIs, and deployment tools like Ollama and LangServe.
• Includes advanced topics such as multi-agent systems (Swarm, SDK) and human-in-the-loop workflows.

Maintenance & Community

The project is actively updated, with modules 01 through 13 currently available. Contributions via Issues and Pull Requests are encouraged. Technical discussions can be initiated via WeChat (ID: a19731567148, mention "Agent").

Licensing & Compatibility

The provided README content does not specify a software license. Potential users should verify licensing terms for compatibility with commercial use or integration into closed-source projects.

Limitations & Caveats

Modules 14 (Comprehensive Practical Project) and 15 (Deployment and Summary) are marked as "under development" (🚧撰写中), indicating that content for these advanced sections is not yet complete.