What is Git?

Git in One Sentence

Git is a distributed version control system that tracks content as a directed acyclic graph (DAG) of immutable snapshots. Every commit is a pointer to a complete tree of files, and every tree is identified by a SHA-1 hash of its content. If even a single byte changes, the hash changes, which means integrity is built into the data model itself.

Unlike tools that store patches (delta-based systems), Git snapshots the entire working tree at each commit. It then compresses identical objects behind the scenes, so storage stays efficient without requiring deltas.

Distributed vs Centralized

Centralized version control systems (SVN, CVS, Perforce) store the canonical repository on a single server. Developers check out files and must be connected to the server to commit. Git gives every clone the full history — commits, branches, tags, and all metadata. You can commit, branch, and diff locally without any network access.

This distributed architecture means every developer's machine is a full backup of the repository. If the server dies, any clone can restore the entire history.

Content-Addressable Storage

Git stores every piece of data — file content, directory structure, commit metadata — as a named object in .git/objects. The name is the SHA-1 hash of the object's content. This is content-addressable storage: the address (hash) is derived from the content itself.

mkdir my-project && cd my-project
git init
ls .git/objects

echo "Hello, Git" > greeting.txt
git add greeting.txt
git cat-file -t $(git hash-object greeting.txt)

The Three States

Every file in a Git repository exists in one of three states. Understanding these states is the foundation for every Git command you will ever run.

You edit files in the working tree. When a change is ready, you git add it to the staging area (also called the index). When you git commit, the staging area becomes a new repository commit. This three-step workflow — edit, stage, commit — gives you precise control over what goes into each commit.

echo "version 1" > README.md
git status                  # README.md is untracked (working tree only)

git add README.md
git status                  # README.md is staged (in the index)

git commit -m "initial commit"
git status                  # working tree is clean (nothing to commit)

What Git Ignores

Not everything belongs in version control. Build artifacts, dependencies, OS files, and secrets should stay out of the repository. Git uses .gitignore to specify patterns for files and directories it should never track.

# Dependencies
node_modules/
vendor/

# Build output
dist/
build/
*.o
*.exe

# OS files
.DS_Store
Thumbs.db

# Secrets
.env
*.pem