AI Agents for Beginners

What it does

• Eighteen lessons (as of this review): intro and agentic design patterns through tool use, planning, agentic RAG, multi-agent, metacognition, production and securing agents
• Later lessons track the current agenda: agentic protocols (MCP, A2A, NLWeb), context engineering, agent memory, Microsoft Agent Framework, computer-use agents, local agents
• Every lesson pairs a written README and a short video with Python code samples built on Microsoft Agent Framework and Microsoft Foundry
• 50+ community translations via Co-op Translator, with a sparse-checkout recipe to clone without the translations bulk

When to use it

The structured on-ramp for developers new to agents, and a refresher map for experienced ones — the lesson table doubles as a curated index into design patterns and the current protocol and tooling landscape.

Getting started

# browse on GitHub, or clone; the repo is large because of translations
# sparse clone without translations (from the README):
git clone --filter=blob:none --sparse https://github.com/microsoft/ai-agents-for-beginners.git
cd ai-agents-for-beginners
git sparse-checkout set --no-cone '/*' '!translations' '!translated_images'

Things to know

Code samples are Python on Microsoft Agent Framework + Microsoft Foundry — the course is Microsoft-flavored despite the generic title, so port the patterns if you're on another stack. Later lessons (Microsoft Agent Framework onward) are text-first, with no video yet. Video links point at YouTube.