Before you begin

This course assumes you’re comfortable opening a terminal and typing commands. You don’t need to be a software engineer, but a handful of terms will come up on every page, and knowing what they mean turns a wall of jargon into readable prose.

If any of this is already obvious, skip to Installing Claude Code.

Your machine

• Terminal: the window where you type commands at your computer. macOS ships with one called Terminal. It’s just an app, like a browser is an app.
• Shell: the program running inside the terminal that interprets what you type. On modern macOS that’s zsh. You’ll rarely need to care which shell you’re on, but you’ll see the word in places like ~/.zshrc (the config file your shell reads at startup).
• Command line and CLI: synonyms for “type commands at a prompt.” Claude Code is a CLI tool.

Two commands you should know:

• cd some/folder: change directory.
• ls: list what’s in the current directory.

That’s enough to follow every example in this course.

Files, repos, and git

• Repository (repo): a folder on your computer containing a project’s files, tracked by a tool called git. Claude Code works best when it’s run inside a repo, because the repo is the unit Claude reads, edits, and commits to.
• Git: version control. Every change to every file is tracked so you can see what changed, when, and why, and roll back if something breaks. Claude uses git constantly; you’ll see commands like git status, git diff, git commit in the transcript.
• GitHub: a website that hosts git repos. Useful for sharing code, reviewing changes with others, and running automated checks. Not required to work with Claude locally.

If you’ve never used git, install it (brew install git) and skim the official Git handbook intro. You don’t need to master it. Just know that git is how “state” is preserved across sessions.

What an LLM is, in one paragraph

A large language model (LLM) is a program trained on an enormous pile of text. It learned, from that text, the statistical shape of human language well enough to continue any sentence you give it in a plausible, coherent way. Modern LLMs (Claude is one) extend that base capability with tools (the model can ask to run a command, read a file, search the web) and reasoning (the model works through a problem step-by-step before answering). Everything else in this course is mechanics and craft on top of that core capability.

A few words you’ll see constantly:

• Token: the unit the model reads and writes. Roughly 3–4 characters, or about ¾ of a word in English. A 1,000-word essay is ~1,300 tokens. A model’s context window is measured in tokens.
• Context window: how much text the model can “see” in a single request. As of mid-2026, frontier models accept up to about 1 million tokens.
• Prompt: what you send the model. Your question, plus anything else the harness attached (conversation history, system instructions, file contents).
• Harness: the local program that wraps the model. It holds the conversation, decides what to send on each turn, dispatches tool calls, and displays results. Claude Code is a harness.
• API: how programs talk to the model directly, without a harness. Most of this course assumes you’re using Claude Code; the API comes up only when you want to script things yourself.

Markdown

A .md file is just a plain-text file with a few conventions: # for headings, * for bold, - for bullets. Any text editor can open one. Markdown matters because LLMs read it as naturally as English, and because it diffs and versions cleanly in git. Markdown as the medium goes deeper; for now, know that when this course says “write it in markdown,” it means “write it in a .md file.”

What’s next

You have the vocabulary. The next three chapters get you installed and configured. From there, the course is mostly about how to work with Claude well, which is the part that compounds.

• Installing Claude Code: getting the tool on your machine.
• A markdown viewer: a companion app for reading all the .md files you’ll accumulate.
• Configuring Claude: turning on the settings that make Claude Code productive.