Course overview

Most people meet AI through a chat window. You type, it answers, you copy the answer somewhere useful. That works, and it has a ceiling. The ceiling is that the tool never touches your actual files. You are the transport layer between the AI and your work.

Claude Code removes that step. It runs on your computer, reads the folders you point it at, and does the work in place. Anthropic describes it as an agentic tool that reads your files, edits them, and runs commands. The documentation is explicit that it can help with anything you could do from the command line, including writing documents, searching files, and researching topics.

This course gets you from nothing to a working install and one finished real task. It is deliberately not a programming course. Every example is a business job: sorting a folder of contracts, turning a pile of PDFs into a one-page summary, building a report you rebuild every month.

You will learn what Claude Code actually is and how the working loop differs from a chat exchange. You will install it, log in, and understand why it stops to ask permission. You will run the loop yourself: write a prompt, watch it read and act, review what it changed, approve or correct. You will learn how to stop it mid-task, how to rewind a file edit, and how to keep a copy of anything you cannot afford to lose.

Then you will do a real task start to finish, with a second pattern you can reuse. The course closes on judgment: when to open a chat window instead, what this costs in plain terms, the safety habits worth keeping, and where to go next.

This is a Supervise course, the third rung of the Nantucket AI trust ladder. Supervise means the tool does the work and you watch it happen. You are not reviewing a finished draft after the fact. You are present while it runs, and you can stop it. That is the whole skill.

Terminal nervousness is normal and this course expects it. You will type a small number of commands. Nothing you type here can break your computer, and Module 3 covers exactly how to undo what the tool does.

Who this is for

• Business owners and managers who already use AI chat and want more from it
• Nonprofit staff with a filing problem and no budget for software
• Office managers who handle documents, records, and recurring reports
• Anyone comfortable in Claude or ChatGPT who has never opened a terminal
• Teams evaluating whether Claude Code belongs on staff computers

No coding background is required. No prior terminal experience is required.

What you will be able to do

By the end of this course you will be able to:

• Explain what Claude Code is, what surfaces it runs on, and how it differs from a chat window
• Say which Claude plans include Claude Code and which do not
• Install Claude Code on your own machine and confirm the install worked
• Start a session in a specific folder and end it cleanly
• Read a permission prompt and decide whether to approve it
• Write a first prompt specific enough to get a useful result
• Watch a session and recognize what the tool is doing at each step
• Stop a running task, correct it mid-flight, and rewind a file edit
• Protect anything a rewind will not cover
• Complete a real document task end to end without writing code
• Reuse a recurring-report pattern on your own work
• Choose between chat and Claude Code for a given job
• Describe in plain terms what drives the cost of a session
• Name the four configuration surfaces you will learn next

Assessment

Each module ends with a short quiz. The course ends with a final assessment covering all five modules.

Passing grade: 70 percent. Retakes are unlimited. Your score is not published or shared with anyone.

What this course does not cover

Claude Code is configurable. There is a file that gives it standing instructions, a settings file that controls its behavior, a way to package your own repeatable procedures, and a way to enforce rules it cannot talk its way around. All four are real, all four are useful, and none of them are in this course.

That material is The Four Surfaces of Claude Code, the Delegate-level course that follows this one. You do not need any of it to finish a useful job today. Module 5 names each surface so you know what you are looking at when you get there.

Modules

1. What Claude Code Is and Is Not (3 lessons) – The agentic loop against the chat exchange, the surfaces it runs on, and what supervising actually asks of you.
2. Getting Set Up (4 lessons) – Account and plan, installing it, your first launch, and why it stops to ask permission.
3. The Working Loop (4 lessons) – Writing the prompt, watching it work, steering and stopping, and undoing what you did not want.
4. Your First Real Task (3 lessons) – Setting up a workspace you cannot damage, a guided document task, and a recurring-report pattern you can reuse.
5. When to Reach For It Over Chat (4 lessons) – Which tool for which job, cost in plain terms, safety habits at Supervise level, and what the next course covers.

After this course

The next step is The Four Surfaces of Claude Code at the Delegate level. It covers CLAUDE.md, settings.json, skills, and hooks, plus model selection and what it does to your bill. Take this course first. The configuration only makes sense once you have watched the loop run.

Frequently asked questions

Do I need to know how to code?

No. Claude Code was built for software and Anthropic’s own documentation says it can help with anything you could do from the command line, including writing documents, searching files, and researching topics. This course teaches the document side. There is no code in any lesson.

Is the terminal going to break something on my computer?

The terminal is a window that prints text and waits for you to type. A command it does not understand produces an error message and a fresh prompt. That is the entire failure mode. Claude Code itself asks before it edits a file or runs a command, and Module 3 teaches you how to undo file changes and how to work on a copy so the originals are never at risk.

Do I need to pay for something before I start?

Yes. Claude Code is not included in the free Claude plan. It is included in Pro, Max, Team, and Enterprise. Module 2 covers the current plan details and recommends starting with a single seat, proving the value on real work, and only then buying more.

How is this different from just using Claude in a browser?

Chat returns text and you move it where it needs to go. Claude Code opens the real files on your computer, works across hundreds of them at once, and writes the result to disk. The simplest test is in the last module: at the end of this job, what exists that did not exist before? If the answer is a better idea, use chat. If the answer is a file, use Claude Code.

What does it cost to run?

Claude Code charges by API token consumption, and Anthropic documents the drivers plainly. Module 5 covers what runs a bill up, the handful of habits that keep it down, and the /usage command that shows you where you stand. It also explains why the published developer averages are an upper bound rather than an estimate for a shop owner running two sessions a week.

Should I take this or the configuration course first?

This one. The Four Surfaces of Claude Code covers CLAUDE.md, settings.json, skills, and hooks, plus model selection and subagents. All of it solves problems you have not had yet. Take this course, run real sessions for a few weeks, and go to the next one when you notice yourself retyping the same instructions for the third time.

Where this course sits

This is a Supervise course, the third rung of the Nantucket AI trust ladder. It applies the H.E.L.P. method you may already know from The H.E.L.P. Framework. Next: The Four Surfaces of Claude Code.