GitHub without fear: four ideas every AI beginner should know

Four words that explain GitHub

Opening GitHub for the first time can feel like walking into someone else's control room. Branches, requests and actions fill the screen, while every button seems to assume that you have been programming for years. You do not need that background to start using GitHub with AI.

This series begins with a safe practice space. Today you will create it. The next instalment will connect that space to ChatGPT, and a later guide will show you how an AI agent can suggest edits without quietly taking over the working version of your project.

A repository is a project folder stored on GitHub. It can contain software, but it can just as easily hold articles, images, research notes or any other files. Our practice repository needs only a `README.md`, the introductory document GitHub displays on its front page.

Unlike an ordinary folder, a repository remembers its history. You can inspect earlier versions, compare what changed and return to a previous state. That history matters when AI joins the workflow because a questionable edit remains visible. You are reviewing evidence, not trusting a chatbot's confidence.

A commit is a saved snapshot accompanied by a short note. “Add project introduction” will help you later. “Stuff” probably will not. The note should explain what changed well enough for your future self to recognize that point in the history.

A branch is a separate workspace that leaves the main version, usually called `main`, untouched. An AI agent can edit a branch while the working project stays as it was. You can keep a useful result or throw away a failed attempt without untangling the two.

A pull request, often shortened to PR, is where that proposed work is reviewed. It shows which lines were added or removed and gives you a place to discuss them before merging anything into `main`. The name sounds stranger than the job. Think propose, inspect, accept.

The whole route is repository, branch, commit and pull request. GitHub offers far more, but those four ideas are enough to understand the rest of this series and to see what an AI coding agent is doing to your files.

Create your first repository

Sign in to GitHub and choose New repository. Give it an obvious name such as `ai-practice`, choose Private, tick Add a README file, then select Create repository.

That is your safe workspace. You have not launched a website or connected an AI. Open `README.md`, select the pencil icon, add two lines describing the folder and choose Commit changes. A useful message would be “Explain the purpose of this repository”.

GitHub's own beginner tutorial confirms that this browser exercise requires no coding, Git installation or command line. If you want the AI half demystified too, our guide to how ChatGPT works provides the missing context. We also have a plain-English explanation of what AI agents are and how they act.

Next time, you will connect this single practice repository to ChatGPT with read-only access. The AI will be able to find and explain files, while your content remains out of reach of its editing tools.