8.8K ·GitHub
What is this?
Agent Learning Hub is a curated, stage-by-stage roadmap and resource library for learning to build AI agents, covering everything from a minimal tool-call loop to shipping a production-ready multi-agent system.
Agent Learning Hub is a curated, stage-by-stage roadmap and resource library for learning to build AI agents, covering everything from a minimal tool-call loop to shipping a production-ready multi-agent system.Is this a framework to install, or just a reading list?It is a single README organized as a checklist-style curriculum with linked articles, papers, and open-source project references — there is no code or framework to install here.
It is a single README organized as a checklist-style curriculum with linked articles, papers, and open-source project references — there is no code or framework to install here.Where should a beginner start?Start at Stage 0 by reading Anthropic's 'Building Effective Agents' and OpenAI's 'A practical guide to building agents', then write a short note on why your use case needs an agent before moving to Stage 1's minimal agent loop.
Start at Stage 0 by reading Anthropic's 'Building Effective Agents' and OpenAI's 'A practical guide to building agents', then write a short note on why your use case needs an agent before moving to Stage 1's minimal agent loop.Does it include working agent code I can run?No — it is a curriculum of checklists and links to external guides, papers, and open-source projects; you build and run the actual agents yourself using the referenced resources at each stage.
No — it is a curriculum of checklists and links to external guides, papers, and open-source projects; you build and run the actual agents yourself using the referenced resources at each stage.How is the roadmap organized?It runs through 8 stages, from building a minimal agent loop and RAG/memory basics to studying a real agent harness, multi-agent coordination, skills/protocols, browser agents, evaluation, and finally shipping a real agent.
It runs through 8 stages, from building a minimal agent loop and RAG/memory basics to studying a real agent harness, multi-agent coordination, skills/protocols, browser agents, evaluation, and finally shipping a real agent.What is the Project Ladder?It is a table of 11 progressively harder hands-on projects, from a simple Calculator Agent up to a Production Harness, each naming the specific skill it teaches such as RAG, browser automation, or multi-agent writing.
It is a table of 11 progressively harder hands-on projects, from a simple Calculator Agent up to a Production Harness, each naming the specific skill it teaches such as RAG, browser automation, or multi-agent writing.Which agent harnesses does Stage 3 recommend studying in depth?It points to Claude Code, learn-claude-code, claw0, hello-agents, OpenClaw, Hermes Agent, CyberClaw, and LangGraph, chosen to study how real systems organize tools, context, permissions, and state.
It points to Claude Code, learn-claude-code, claw0, hello-agents, OpenClaw, Hermes Agent, CyberClaw, and LangGraph, chosen to study how real systems organize tools, context, permissions, and state.What are similar projects?awesome-chatgpt-zh, open-multi-agent, ai-system-design, ruflo, awesome-agentic-ai-zh, ai-guide, awesome-ai-harness, hive
awesome-chatgpt-zh, open-multi-agent, ai-system-design, ruflo, awesome-agentic-ai-zh, ai-guide, awesome-ai-harness, hive