AI Agent Engineer

AI Agent Engineer

Git Cheat Sheet: Getting Started and Creating a Repository

3 min read

Getting Started and Creating a Repository

This command group starts a Git repository: either initialize the current directory or copy an existing remote repository locally. This page recognizes the two entry points first. Before staging and committing, continue to Git State Model and Commit Graph to build a mental model of snapshots, index, HEAD, and branches.

git init

git init

Purpose: create a new Git repository in the current directory. It creates a .git/ directory where Git stores commit history, branch references, configuration, and staging-area state.

Appropriate when:

  • the current directory already contains files but is not yet managed by Git;
  • you want to start a local project from scratch before deciding whether to add a remote.

Notes:

  • Run it at a project root, not accidentally in an unrelated parent directory.
  • Re-running it in an existing Git repository normally does not destroy history, but is unnecessary.
  • git init creates only a local repository; it does not automatically connect GitHub, GitLab, or another remote.
  • The default branch name depends on Git version, installer, and init.defaultBranch configuration. When an experiment needs a definite name, run:
git init --initial-branch=main

Git 2.55’s built-in default remains master, and the official plan is to switch it to main in Git 3.0. An environment may already override that, so always gather evidence with git branch --show-current.

Common next commands:

git status
git branch --show-current
git remote

At this stage, observe only. Staging, committing, and pushing are covered later; do not treat git add ., commit, and push as a fixed initialization bundle when scope is uncertain.

git clone <url>

git clone https://github.com/example/project.git

Purpose: copy a complete project from a remote repository, including file contents, commit history, branch references, and remote-address configuration.

Appropriate when:

  • you already have a remote repository address, such as a GitHub repository URL;
  • you want to develop, read, or run an existing project locally.

Common forms:

git clone https://github.com/example/project.git
git clone git@github.com:example/project.git

Notes:

  • HTTPS addresses are usually easier to use directly; SSH addresses require an SSH key first.
  • The initial cloned branch is set by remote HEAD, so do not assume it is always main.
  • Git creates a directory matching the repository name by default. To choose a local directory name:
git clone https://github.com/example/project.git project-local

Verify with:

git status
git remote -v

git status confirms working-tree state; git remote -v confirms remote addresses.

Remote addresses can leak secrets

Teaching examples use public fictional addresses. Do not put a token, password, or userinfo credential in a URL. Before publishing git remote -v output, inspect and redact it.

Read-only observation exercise

In an existing repository whose purpose you clearly know, run only:

git rev-parse --show-toplevel
git status --short --branch
git branch --show-current
git log -1 --oneline --decorate
git remote

Answer: where is the repository root, what is the current branch, is the working tree clean, what is the latest commit, and which remote names are configured? Do not run init, add, commit, or push in a real repository merely to practice. For write operations, enter the isolated integrated project.

References

Checked against Git 2.55.0 official documentation, obtained 2026-07-14.

Backlinks