Code-Agents-Playbook

> A playbook for learning how coding agents are put together — and for building your own

The playbook is built on examples from real production coding agents, including Claude Code. 18 chapters bring together best practices in architecture, patterns, and production trade-offs into a reference you can act on.

Two ways to use it:

1. 📖 Study — Read the chapters to understand how a production-grade code agent actually works under the hood
2. 🛠️ Build — Hand this playbook to an AI agent (or follow it yourself) as a blueprint to create your own code agent from scratch

What's inside:

• 📝 18 chapters of original prose and diagrams covering every subsystem — from the agent loop to IDE integration
• 🐍 Runnable Python code samples (code-samples/) — original implementations that demonstrate each pattern
• ✅ "Build your own" checklists in every chapter — step-by-step guidance to reimplement each subsystem

Chapter structure (same across all parts — from intuition to detail):

| Section | What you get | | ------------------------ | ----------------------------------------------------------------------- | | Overview | What the subsystem does and why it matters, in plain language. | | How it fits together | Architecture, data flow, and mermaid diagrams — how the pieces connect. | | Production concepts | Practices and mechanisms inferred from production coding agents. | | Key design decisions | Trade-offs and non-obvious choices. | | Insights | Gotchas, edge cases, and lessons worth remembering. | | Code samples | Links to runnable Python in code-samples/ for that chapter. | | Build your own | Concrete steps to reuse the pattern in