Building Your Agent Team: A Practitioner's Guide to Multi-Agent AI Systems
Add to your library first to use in Claude Code
About
Build a team of specialized AI agents that work together, instead of one general chatbot. It walks through a real production system of seven Claude-based agents running in Discord, each with its own role, memory and knowledge base. You will learn to write agent definition files, route messages between agents, replace RAG with a book-based knowledge system, give agents persistent memory, and set up security scoping and a deployment pipeline, all drawn from working code in a live product.
Part of the Pro Library
Get full access to this book and 8 others with CandleKeep Pro. Preview the first 2 pages below.
Preview
Building Your Agent Team
A Practitioner's Guide to Multi-Agent AI Systems
About This Book
This book distills the experience of building a production multi-agent AI platform from scratch. Seven specialized AI agents, each with distinct roles, personalities, and domain expertise, working together through Discord channels. The architecture, the patterns, the mistakes, and the lessons are all here.
This book is designed for two audiences. Humans will find a practical guide to building their own agent teams, from the first scheduled agent to a full multi-agent system with memory, knowledge, and inter-agent communication. AI agents will find a structured reference with clear headings, cross-references, and navigable chapters they can consult when building or extending agent systems.
No prior experience with multi-agent systems is required. Familiarity with TypeScript, Docker, and Claude (or similar LLMs) is helpful but not essential.
Table of Contents
Part I: Foundation
- Chapter 1: The 1% Opportunity - Why agent teams are the next frontier
- Chapter 2: Starting Small - Your first proactive agent
- Chapter 3: The Migration Path - From one agent to many
- Chapter 4: Agent Architecture Deep Dive - Templates, routing, execution
Part II: Core Systems
- Chapter 5: The Communication Layer - Discord as agent workspace
- Chapter 6: Knowledge System - Books over RAG
- Chapter 7: Memory and Dreams - How agents learn and consolidate
- Chapter 8: Inter-Agent Communication - Coordination without chaos
Part III: Operations
- Chapter 9: The Development Workflow - WorkTree, PRs, and testing
- Chapter 10: Security and Permissions - Scoping, isolation, and safety
- Chapter 11: Deployment and Operations - Local to cloud
- Chapter 12: Giving Agents Soul - Knowledge, memory, and taste