oreilly-ai-agents

This repository hosts a comprehensive collection of AI agent tutorials and notebooks focused on building and evaluating modern automated AI agents using frameworks like CrewAI, LangChain, Autogen, and other agent tooling ecosystems. It is not centered around a running MCP server, but you can leverage the included notebooks and example workflows to prototype and test agent behaviors, tool usage patterns, and multi-step task execution. Users typically explore how agents select and invoke tools, manage prompts, and reason through tasks, then adapt these patterns to their own MCP-enabled environments as needed. The materials help you understand agent orchestration, tool integration, and evaluation techniques across multiple frameworks.

To make practical use of the content, start with the notebooks and hands-on examples that demonstrate ReAct-style agents, multi-agent coordination, RAG workflows, and tool selection strategies. You’ll find references to LangGraph, SmolAgents, CrewAI, Autogen, OpenAI tools, and LangGraph-based MCP integrations in the materials. Use these resources to design agent-powered tasks, run experiments locally, and iterate on tool selection, prompting, and execution flows before porting concepts into an MCP server setup if required for your project.

Prerequisites

• Python 3.11 installed on your system
• Git to clone the repository
• Optional: a virtual environment tool (venv is included in Python)

Installation steps

1. Clone the repository

git clone https://github.com/sinanuozdemir/oreilly-ai-agents.git
cd oreilly-ai-agents

2. Set up a Python 3.11 virtual environment

• macOS/Linux

python3.11 -m venv .venv
source .venv/bin/activate

• Windows

python3.11 -m venv .venv
 .venv\Scripts\activate

3. Install Python dependencies (if a requirements.txt is present at the repo root or within notebooks folders)

pip install -r requirements.txt

4. Start using the notebooks and examples

• You can launch Jupyter Notebook or JupyterLab inside the activated environment

python -m jupyter notebook

5. (Optional) Install additional tooling referenced in the notebooks (e.g., LangGraph, CrewAI, Autogen) following their respective installation guides.

Notes

• Some examples may require API keys or access to external services (OpenAI, etc.). Configure environment variables as instructed in the individual notebooks.

Tips and considerations:

• Work within a Python 3.11 virtual environment to avoid dependency conflicts.
• The repository content spans multiple agent frameworks; ensure you install the specific dependencies mentioned in each notebook's setup cells or README notes.
• If you plan to run MCP-related integrations later, you can adapt the demonstrated tooling patterns (tool calls, prompts, and ReAct-style reasoning) into your MCP server workflow.
• Common issues often involve API key configurations, version mismatches between frameworks, or missing runtime dependencies for notebooks (Jupyter, streamlit). Check notebook comments and setup sections for guidance.
• For experimentation, start with small, well-defined tasks in notebooks to validate tool calls and state management before scaling to larger agent coordination scenarios.