The Context Engineering Handbook

Chapter 1. The Road from Prompt Engineering to Context Engineering (and Where We Go From Here)

By Mike Taylor

A Note for Early Release Readers

With Early Release ebooks, you get books in their earliest form—the author’s raw and unedited content as they write—so you can take advantage of these technologies long before the official release of these titles.

This will be the 1st chapter of the final book. Please note that the GitHub repo will be made active later on.

If you’d like to be actively involved in reviewing and commenting on this draft, please reach out to the editor at [e-mail].

This chapter will deliver a brief history of “prompt engineering” as a practice, explaining how it evolved around single-turn LLM applications and chatbot usage, and connect it to the emergence of “context engineering”. The chapter is based on my experience as an AI engineer from 2020 to today, including anecdotes of what problem each new evolution solved. I’ll introduce you to the technologies and patterns that gave rise to “context engineering”, and how it differs from prompt engineering. We’ll explore a few potential futures, including prompt optimization and managing fleets of agents, and why we’ll never escape the need to clearly communicate what we want AI programs to do.

Origins and Early Days (2018–2021)

The transformer model was conceived in 2017 with the famous “Attention Is All You Need” paper, but GPT-2 in 2019 was the first language model that I saw smart people in my ...