All commands

git init

Turn any folder into a Git repository.

git init
Where this fits in the command map

What it does

git init creates a hidden .git directory inside your project. That directory is the entire repository — every object, every pointer, every piece of history lives inside it. From this moment, Git can track what happens to every file in the project.

your-project/ · after git init
.git
├── HEAD
├── index
├── objects/
│ ├── info/
│ └── pack/
└── refs/
├── heads/
└── tags/

git init materializes the .git folder

Watch it — press play

A narrated, step-by-step walkthrough that shows how to do it — not just theory. It adapts to the learner mode you picked in the top bar.

1 / 3

Step 1

Choose an empty room

Point git at any folder — even one with nothing in it yet.

git init
your-project/ · after git init
.git
├── HEAD
├── index
├── objects/
│ ├── info/
│ └── pack/
└── refs/
├── heads/
└── tags/

A bare folder, before git arrives

Under the hood

1

Git creates the .git folder

git init writes a .git directory containing the four core structures: objects/ (empty), refs/ (empty), HEAD (a file pointing at refs/heads/main), and index (an empty binary staging file).

mkdir .git
mkdir .git/objects/{info,pack}
mkdir .git/refs/{heads,tags}
echo 'ref: refs/heads/main' > .git/HEAD
2

HEAD is born

HEAD is just a text file holding the name of the current branch. It points at refs/heads/main, but that branch file doesn't exist yet — it appears with your very first commit.

cat .git/HEAD  # → ref: refs/heads/main
3

Nothing is tracked yet

No objects exist. No files are in the index. Git simply says 'demonstrated' and waits. The working directory itself is untouched — Git is a layer on top, not a copy.

Command anatomy

git init [<directory>]
<directory>

Initialize a repo in a specific folder instead of the current one.

-b <name>

Set the initial branch name (defaults to main).

--bare

Create a bare repo — no working directory, refs only. Used on servers.

When to reach for it

Starting a brand-new project and you want version control from day one.

You downloaded someone's code without a .git folder and want your own history.

You're exploring Git internals and want a clean sandbox: git init /tmp/lab.

Practice

A real repo, pre-seeded with history, runs right here — stage, commit, branch, stash, diff, even hit and resolve a real merge conflict. Nothing to install, nothing leaves your machine.

Open practice playground

Go deeper with

git add

Copy your edits into the staging area (the index).

git commit

Seal the index into a permanent commit object.

git clone

Copy an entire repository — history, refs, branches — to your machine.