AI Agent Book is an open-source textbook and experiment collection focused on the design principles and engineering practice of AI agents. Built around the formula “Agent = LLM + context + tools,” it develops the subject across ten chapters and pairs the written material with dozens of runnable experiments.

The book progresses from foundational agent concepts into context engineering, tool use, evaluation, training, graphical interfaces, robotics-related experiments, and multi-agent architectures. The repository provides PDF and EPUB builds, source chapters, reproducible environments, and community translations into numerous languages.

It is designed for readers who want both conceptual explanations and code they can run while learning. Rather than presenting one agent framework as the answer, the project exposes the underlying components and engineering trade-offs so students and practitioners can understand how modern agent systems are assembled.