Agentic Coding HandbookBuild software with AI agents

A practical guide to agentic coding and AI-assisted development — from first principles to production workflows.

Understand the shift

Move from prompting for snippets to delegating whole features. Learn how the agentic loop — understand, plan, implement, verify — actually works.

Learn the paradigm

Install Claude Code, wire up your project with CLAUDE.md, and run your first agentic task in about five minutes.

Get set up

Scope, constraints, examples, and layered iteration — the difference between an agent that flails and one that ships.

Write better prompts

Keep the agent focused as sessions grow long, and break big work into reviewable, verifiable steps.

Stay in control

Make the agent prove its own work with tests and types, so your review focuses on design, not mechanical bugs.

Close the loop

Claude Code commands, CLAUDE.md structure, hooks, and MCP — the details, kept easy to keep current.

Look things up