Summary

This repository provides a curated roadmap and resource hub for learning AI Agent development, targeting individuals from beginners to experienced engineers. It aims to guide users in building reliable, production-ready agents by focusing on practical skills and modern architectures, rather than just collecting links.

How It Works

The hub structures learning through progressive stages, from understanding agent fundamentals to implementing advanced concepts like tool use, RAG, memory, multi-agent coordination, and skills. It prioritizes current, production-oriented directions such as coding agents and harness engineering, offering a clear learning path and project ladder for hands-on experience.

Quick Start & Requirements

This repository serves as a learning guide, not a direct software installation. Users are expected to follow the "Learning Todo List" and "Project Ladder" stages. Prerequisites for building agents include access to LLM APIs, Python environments, and potentially specific libraries like LangChain or LlamaIndex, as detailed in the "Curated Resources" section. Links to official documentation and example projects are provided for practical implementation.

Highlighted Details

• Current Learning Priorities: Focus on Claude Code/Codex-style coding agents, agent harness engineering, personal agents (OpenClaw/Hermes), skills/protocols (MCP/A2A/ACP), and evaluation/safety.
• Structured Learning Path: Stages 0-8 cover agent basics, minimal loops, tool use, RAG, memory, modern harnesses, multi-agent systems, skills, browser/computer use, and evaluation.
• Project Ladder: A 11-level progression from simple "Calculator Agent" to "Production Harness" for practical skill development.
• Extensive Resources: Curated links to official guides (Anthropic, OpenAI, Google), research papers, and numerous GitHub repositories for hands-on learning.

Maintenance & Community

The hub is curated by 陈思州 (Datawhale 成员). No specific community channels (e.g., Discord, Slack) or detailed contribution guidelines beyond general preferences are listed.

Licensing & Compatibility

The repository's README does not specify a software license. This omission requires users to investigate potential licensing implications for any associated code or frameworks referenced.

Limitations & Caveats

The repository explicitly advises against prioritizing older, "role-play" agent frameworks, recommending a focus on current, production-oriented directions. The rapid evolution of the AI agent field means the recommended learning path may require continuous updates.