Repository, Git Object, Branch

Local and Remote Repositories

Local Repository
= A Git repository on your computer

Remote Repository
= A repository on GitHub, GitLab, or an internal Git server

One local repository can have multiple remotes configured.

Local Repository
├─ origin  → GitHub
├─ backup  → GitLab
└─ upstream → Official source repository

What Is a Git Object?

Git stores data internally as objects.

Object	Role
Blob	File contents
Tree	Filenames and directory structure
Commit	A snapshot at a particular point in time + metadata
Tag object	An annotated tag attached to a specific object

flowchart TD
    C[Commit] --> T[Root Tree]
    T --> B1[Blob<br/>Camera.cpp contents]
    T --> B2[Blob<br/>Scene.cpp contents]
    T --> ST[Sub Tree]
    ST --> B3[Blob]

A commit typically contains the following information.

Commit
├─ Tree
├─ Parent Commit
├─ Author
├─ Committer
├─ Message
└─ Time

A Branch Is Not an Object

A branch is a movable reference pointing to a particular commit.

feature/camera
      ↓
   Commit D

Creating a new commit moves the branch to that commit.

Before
A ─ B ─ C
        ↑ feature

After commit
A ─ B ─ C ─ D
            ↑ feature

Bare Repository

A bare repository contains Git data without a working tree.

Regular repository:

Project/
├─ Source/
├─ README.md
└─ .git/

Bare Repository:

project.git/
├─ objects/
├─ refs/
├─ HEAD
└─ config

A bare layout is natural for a server because the server does not need to edit files directly.

Reasons to Use Multiple Remote Repositories

Backups
GitHub + GitLab mirrors
Separating a fork from the official source
Separating internal and external distribution
Separating permission boundaries

However, there is no need to split remote repositories merely because Claude and Codex develop in parallel. In that case, one remote with multiple branches/worktrees is simpler.