Who This Book Is For (and Who It Isn't)

Draft v0. A living starting point — the three “is not” points the author named are in here, plus several that follow from the book’s positioning. We’ll cut, sharpen, and reorder once the voice settles.

Most books open by trying to convince you they’re for everyone. This one does the opposite. It is a narrow book on purpose. The fastest way to know whether it will be worth your time is to read the two lists below and see which one you land in. If you land in the second list, put the book down with no hard feelings — you’ll have saved yourself an afternoon, and I’d rather you find that out on this page than three chapters in.

A book that tries to serve everyone serves no one well. This one picks its reader.

This book is for you if…

• You’re already using Claude to write code, and it’s mostly working — until it isn’t. You can prompt fluently. You’ve gotten prototypes running. Maybe you’ve shipped something. But every so often Claude leads you somewhere that feels wrong, and you can’t quite name what went sideways or how to get back.

• You’re a non-developer who wants to stop getting surprised. You don’t know what an API is, or what JSON is, or why the command Claude gave you won’t run on your machine — and you’re tired of nodding along. You don’t want to become a programmer. You want enough footing to tell when Claude is about to cost you a weekend.

• You’re a working developer who hasn’t gone deep on Claude Code yet. You can read code. You don’t need anyone to explain what an API is. What you want is an honest field report — what actually goes wrong in real use — from someone who has shipped with it, not a marketing tour.

• You’re a tech lead or manager weighing this for a team. You want to know what changes about code review, what the failure modes are, and what guardrails to put up before you hand this to ten engineers.

• You suspect the experienced engineer matters more in this era, not less — and you want that argument made well. This book makes it.

This book is NOT for you if…

• You want to learn how to make money selling AI consulting services. This is not a get-rich, build-an-agency, “monetize your AI skills” book. There is a whole genre of that, and this isn’t it. Nothing here is about selling — it’s about building software that holds up.

• You want a tour of every Claude Code feature. This is not a reference manual. I won’t walk you through every flag, every slash command, every setting. The book is about patterns — the recurring ways Claude steers you wrong and how to steer back — not a feature catalog. If you want the full feature inventory, the official docs already do that job, and do it well.

• You want a comparison of Claude Code against all its alternatives. This is not a “Claude Code vs. Cursor vs. Copilot vs. everything” survey. I picked Claude Code and went deep. The judgment in these pages transfers to other tools, but I’m not going to pretend to referee a tool war I’m not running.

• You’ve never used an AI coding tool at all. This book assumes you’ve already prompted your way to something. If you’re at absolute zero, start with one of the gentle on-ramps named in the introduction, then come back.

• You want cheerleading about how “anyone can code now.” This book argues close to the opposite. If you want reassurance that the hard parts have disappeared, you’ll find this book irritating.

• You need enterprise governance, compliance, and big-bank scale. Security and cost are covered here for small teams and solo builders — not for a regulated thousand-engineer org. There are good books for that world; this isn’t one of them.

What you will learn

• To recognize Claude’s recurring mistakes by name. The book names a catalog of patterns — the specific, repeatable ways Claude goes wrong — so you can spot one while it’s happening instead of after it’s cost you.

• Why each mistake happens. Not just “watch out for this,” but the mechanism underneath it — so the recognition sticks and transfers to cases I never thought to list.

• The recovery move for each pattern. What to actually say, ask, or do when a pattern fires. Concrete prompts and escalation moves, not vague advice to “be careful.”

• How Claude defends its own mistakes — the rhetorical moves it makes when you push back, and how not to get talked out of a correct instinct.

• The production realities nobody markets — tech debt in an AI-written codebase, testing when Claude writes the tests too, what stays a human’s job.

• (For non-developers) enough vocabulary to follow all of the above — a curated foundation, not a programming course. Developers can skip it.

What this book will NOT cover

• How to make money with AI skills. (See above. Not the subject.)
• Every feature of Claude Code. The patterns, not the manual.
• A survey of competing tools. One tool, deep.
• Installation and first-run setup. Other resources do this well and are named in the introduction.
• How to program from scratch. This is not a coding course.
• Enterprise-scale security, compliance, and governance. Small-team scale only.

Found something wrong, unclear, or plainly disagreeable? Open an issue