Learning to code with a powerful AI agent nearby can still leave a beginner stalled: the code arrives faster than the understanding, one small error stops everything, and it is hard to know what is safe to run. This book gives you one focused weekend project and a clear, safe job for Codex at every step, while you read, verify, and decide, so you build real skill instead of pasted code you cannot explain.
The approach is a repeatable agent loop, not blind trust. You will learn what Codex really does, how it reads a repository, proposes changes, and waits for your approval, then set up a workspace with Git, Node, and the Codex tools, and practise a review ritual of reading transcripts, inspecting diffs, and committing at checkpoints. Safety comes first throughout: small prompts, checked outputs, and no command you do not understand.
The book is organized around one complete build, the Minute Ledger, a small app that logs focus sessions and shows daily totals, a seven-day chart, and a streak. Each chapter adds one visible piece and ends with a checkpoint you can test, moving from how programs think, through planning and building, to storage, testing, debugging, review, and a polished finish you can publish and keep using.
Inside, you will find:
• What Codex really does, and how to work with an agent safely
• A complete workspace setup with Git, Node, and your first safe session
• How programs think, taught through the project you are building
• A written plan for the Minute Ledger, with a data model and file map
• Building the data layer, the page, and responses you can see working
• Saving data so it survives, with streaks, weekly totals, and real tests
• Debugging without guessing, a hardening review, and publishing your finished app
This book is for adults with little or no coding experience who are comfortable following careful instructions and curious about working with an AI coding agent, especially anyone who tried generated code before and felt lost when it broke. You need no framework or prior training, only a computer and a weekend. By Sunday you will have a working app, a tested codebase, and a workflow you can trust.
A weekend will not make you an expert, and this book never pretends otherwise. It will leave you with something more useful: a finished project you understand, the confidence to question every line an agent writes, and a clear thirty-day plan for what to practise next. Start Saturday morning, follow the checkpoints, and finish with code that is truly yours. An independent guide, not affiliated with or endorsed by OpenAI.
