sinanuozdemir/oreilly-ai-agents

python3.11 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Usage

• •python3 -m jupyter notebook (run notebooks in the activated virtualenv)
• •Notebooks cover 3rd-party frameworks (SmolAgents, Google ADK, CrewAI, Autogen), OpenAI Swarm / Agents SDK, LangGraph (RAG, ReAct, MCP integration, middleware), agent evaluation, and multi-agent patterns
• •Includes a 'Computer Use with Reasoning LLMs' notebook that lets a reasoning LLM control the local machine (flagged with a warning)

Key features

• ✓Comprehensive guided material for understanding, implementing, and managing AI agents from prototype to production
• ✓Hands-on notebooks for CrewAI, LangChain/LangGraph, AutoGen, OpenAI Agents SDK, Swarm, SmolAgents, and Google ADK
• ✓Sections on evaluating agents (rubrics, alignment, tool-selection positional bias), multi-agent systems (supervisor, dynamic network, parallel nodes), and modern paradigms (Plan & Execute, Reflection)
• ✓Includes building a custom agent framework (Squad Goals) and multimodal/computer-use examples
• ✓A dedicated notebook connects a ReAct agent to an MCP server for tool use

Best for

Learners following the O'Reilly AI Agents course who want runnable agent-framework examples and evaluation tooling.

Caveats

• ⚠Not a runnable MCP server — it is course code / educational notebooks
• ⚠Requires Python 3.11 and installing dependencies via requirements.txt
• ⚠Computer-use notebooks can let an AI operate your local machine (explicit warnings in README)