Agent Systems Handbook by Prompthon
A field guide to building and operating production AI-agent systems, covering agentic workflows, memory, context engineering, MCP and A2A interoperability, evaluation and multi-agent architecture.
The Agent Systems Handbook is a field guide for students, practitioners and builders working on production AI-agent systems, aimed at what makes an agent system production-ready rather than just an easy demo.
It covers agent-system foundations and mental models, agentic workflows such as planning, reflection, tool use and function calling, agent memory and retrieval, context engineering, interoperability through the Model Context Protocol and Agent2Agent, and evaluation, observability and multi-agent architecture.
Its content is produced through an AI-native workflow that combines AI-assisted drafting and synthesis with expert review, and it is organized so a reader can pursue broad understanding, applying AI to daily work, or building real agent systems, without forcing one path.