You're reading from Building Agents with OpenAI Agents SDK Create practical AI agents and agentic systems through hands-on projects

Product type Paperback

Published in Oct 2025

Publisher Packt

ISBN-13 9781806112012

Length 276 pages

Edition 1st Edition

Languages

Python

Tools

OpenAI Python library

Concepts

GPT/LLMs

Author (1):

Habib

View More author details

Table of Contents (15) Chapters

Preface

1. Part 1: AI Agents

2. Introduction to AI Agents FREE CHAPTER

3. Introduction to OpenAI Agents SDK

4. Environment Setup and Developing Your First Agent

5. Part 2: OpenAI Agents SDK

6. Agent Tools and MCPs

7. Memory and Knowledge

8. Multi-Agent Systems and Handoffs

9. Model and Context Management

10. Agent System Management

11. Part 3: Build AI Agents

12. Building AI Agents and Agentic Systems

13. Other Books You May Enjoy

14. Index

MCP

Agentic AI development is still new but is significantly gaining in popularity. As developers started to build more agents and tools, there was no standardized way to connect tools to agents; every SDK did it differently. We are already familiar with how the Agents SDK connects tools to agents, but this differs significantly from other frameworks such as LangGraph or CrewAI. This was difficult – if you built an amazing tool in the Agents SDK, it was difficult to port it to other frameworks. Additionally, it did not make sense to develop a tool that did X, if someone else had already created a tool that did X for another framework.

As a result, a standard protocol was needed to define the connection between tools and agents that could apply to all SDKs, and MCP was born.

What is MCP?

MCP is a standardized protocol that defines how an AI agent can discover and call tools hosted locally or on external servers. Think of MCP as a universal adapter or a USB-C port...