# Understanding AI Agents: Design Principles and Engineering Practice

An open-source Chinese-language book on AI agent design and engineering practice, organized around the formula "agent equals a language model plus context plus tools," with ten chapters and 109 accompanying runnable experiments.

Chapters progress from agent history and classic paradigms, through context engineering (KV cache, prompting, Agent Skills, context compression), user memory and knowledge bases (RAG, structured indexes, knowledge graphs), tools and the MCP protocol, coding agents versus general-purpose agents, expanding an agent's observation and action space (asynchronous interaction, voice, computer use, robotics), agent evaluation, model post-training, continual agent evolution, and multi-agent collaboration.

The text and every experiment are open source, downloadable as PDF or EPUB and readable online with multi-language switching, chapter folding, highlighting and notes, alongside a companion volume on AI infrastructure. Community volunteers have translated the book into more than a dozen additional languages besides the Chinese original.

Python, Apache-2.0 licensed.

Website: https://github.com/bojieli/ai-agent-book