Mindket
Tool guideGitHub Copilot by GitHubFree plan

How to use GitHub Copilot: set up VS Code and start with agent mode

GitHub Copilot is GitHub's AI coding assistant. It completes code as you type, answers questions about your project and, in agent mode, plans a task, edits files and runs commands while you approve each action. It works in VS Code, JetBrains IDEs, Visual Studio and on GitHub.com. Sign in with a GitHub account, open a folder in VS Code and start a chat.

Good for
  • Code suggestions as you type
  • Questions about your code
  • Agent edits in VS Code
  • Issue to pull request
  • Reviewing pull requests

At a glance

  • GitHub Copilot is GitHub's AI assistant, built into editors, GitHub.com and the terminal.
  • In an editor, agent mode edits across files while you approve actions. On GitHub, the cloud agent takes an issue and hands back a pull request.
  • Copilot Free costs nothing, but the cloud agent is part of the paid plans.
  • Pending edits sit in a list you keep or undo, and terminal changes show only in Source Control.
All tools
On this page
  1. YouTube tutorials
  2. What GitHub Copilot is
  3. What GitHub Copilot is good for
  4. When to use GitHub Copilot, and when to pick another tool
  5. How to use GitHub Copilot, step by step
  6. An example to try
  7. Tips for better results
  8. The plans in plain words
  9. Learn more about GitHub Copilot
  10. Frequently asked questions
  11. Sources

What GitHub Copilot is

GitHub Copilot is an AI assistant from GitHub, the code hosting site owned by Microsoft, that helps you write, understand and ship software. It suggests code as you type, explains a codebase, reviews changes and works on tasks you assign it. GitHub announced it on June 29, 2021, and it has grown from autocomplete into agents.

It runs in several places. VS Code, Visual Studio, JetBrains IDEs, Neovim and Eclipse load it as an extension or plugin. On GitHub.com and in the GitHub Mobile app it needs no install. A command-line version and a desktop app exist too, and every plan includes both. Copilot Free costs nothing and has limited features, and paid plans add more models, more agents and more AI credits each month.

What GitHub Copilot is good for

  • Suggesting code as you type: it can finish a line or a whole function from a comment and the code around it. Copilot Free is limited to 2,000 inline suggestions a month.
  • Answering questions about code: ask what a file does or why a test fails in Chat, in your editor, on GitHub.com, in GitHub Mobile or in Windows Terminal.
  • Building and editing across files in agent mode: describe a small app or a fix, and the agent plans the task, edits files, runs commands and checks the result while you approve each step.
  • Turning an issue into a pull request: assign a GitHub issue to Copilot and its cloud agent works in the background, then asks you to review the pull request. It needs a paid plan.
  • Reviewing pull requests: Copilot can summarize a pull request and suggest improvements to it.
  • Letting you choose the model: paid plans let you pick the model for a chat, while Copilot Free and Copilot Student pick one for you automatically.

When to use GitHub Copilot, and when to pick another tool

GitHub Copilot fits people who already work in an editor or on GitHub. The tools in the right-hand column have their own pages. Cursor is an editor built around its agent, Claude Code works from a terminal window, and Ollama and LM Studio run models on your own computer.

Your jobUse Copilot?Or try
Finish lines and functions as you type in VS Code or JetBrainsYes, this is where it startedCursor, with its Tab completion
Change several files from one request, approving each actionYes, in agent modeCursor or Claude Code
Hand a GitHub issue to an agent and get a pull request backYes, on a paid planStay with Copilot, it lives inside GitHub
Work from a terminal windowYes, with Copilot CLIClaude Code
Run a model on your own computer, offlineNot by default, its models run in the cloudOllama or LM Studio
Chat about anything that is not codeNo, it is built for software workChatGPT, Claude or Gemini

How to use GitHub Copilot, step by step

This route starts in VS Code and ends with edits you have kept or undone. The menu names are the ones the VS Code and GitHub docs use.

  1. Make a GitHub account. Sign up at github.com if you have none. An eligible account with no Copilot plan is enrolled in Copilot Free when it signs in to Copilot.
  2. Install VS Code. Download it from code.visualstudio.com and install it.
  3. Sign in to Copilot. Hover over the Copilot icon in the Status Bar along the bottom, select Use AI Features, pick a sign-in method and follow the prompts.
  4. Open a folder. Create an empty folder, choose File > Open Folder and pick it. If VS Code asks whether you trust the folder, select Manage in the notification, then Trust.
  5. Start a chat. Open the Chat view and select + to begin a new chat.
  6. Pick Agent and manual approval. Set Agent to Agent, so it can edit files and run commands, and Permissions to Manual permissions, so it asks before it acts. Ask and Plan are the modes beside it.
  7. Type your request. Describe the app in plain words and press Enter. When an approval prompt appears in the chat, read what the agent wants to do before you accept.
  8. Keep or undo each edit. Open a file from the list of pending changes above the chat box and select Keep or Undo beside each change. Restore Checkpoint on a request rolls the whole request back.
  1. Sign in to Copilot. Hover over the Copilot icon in the Status Bar along the bottom and select Use AI Features.

    An illustration of VS Code with a gold ring around the Use AI Features button above the Copilot icon in the Status Bar.
  2. Open the Chat view and select + at the top to begin a new chat.

    An illustration of the VS Code Chat view with a gold ring around the plus button at the top right of the Chat header.
  3. Set the mode to Agent and the permissions to Manual permissions, so Copilot can edit files but asks before it acts. Then type your request.

    An illustration of the VS Code Chat input box with gold rings around the Agent and Manual permissions dropdowns.
  4. Read each change Copilot proposes. Select Keep to accept it or Undo to throw it away.

    An illustration of VS Code with gold rings around the Keep and Undo buttons in the pending changes row above the chat box.

AI-made illustrations of GitHub Copilot's screens in VS Code, drawn on 6 November 2026 to match these steps. They are not screenshots, so the real app may look different: follow what you see.

Menus and the desktop app change often, so a button may sit somewhere other than these steps say. The tutorials at the top of this page show the steps on screen.

An example to try

Paste this prompt, the VS Code quickstart's own example, into the chat from step 7. It asks for a small app you can try in a browser.

Create a task list web app in a single index.html file with embedded CSS and JavaScript. Let me add, complete, and delete tasks. Save the tasks in local storage so they persist after a page reload. Use no external libraries.
  • The agent creates one index.html file, and it asks before it does anything your settings do not already allow.
  • Open the file in a browser. You should be able to add a task, mark it complete and delete it.
  • Reload the page. The tasks should still be there, because the prompt asks for local storage.
  • If a step fails, describe what you saw in the chat and ask the agent to fix it. Your own pass through the app is the check that counts.

Tips for better results

  • Begin on Manual permissions and loosen it only in a project you know. Allow all and Autopilot skip confirmation for file edits, terminal commands and outside tool calls.
  • Check the Source Control view after any run that used the terminal. The list of pending changes shows only edits made through the agent's own edit tools.
  • Put the whole job and its limits into an issue before you assign it to the cloud agent. It reads the issue and the comments already there, not comments added later.
  • To try Copilot with no install, open the Copilot home page on GitHub.com, choose a repository in the session scope picker and ask where to start reading.
  • Open the Copilot status dashboard from the Status Bar to see how much of your month is used. One agent request can spend several.

The plans in plain words

Copilot Free is for individuals. It has limited features, automatic model selection and a monthly allowance of AI credits. Copilot Student is free for verified students, and verified teachers and maintainers of popular open source projects may qualify for Copilot Pro at no cost.

Pro, Pro+ and Max are the paid individual plans, each with more AI credits a month than the one before. Business and Enterprise are for organizations, where an administrator assigns seats and sets policies. What each costs today is on GitHub's plans page. The cloud agent belongs to the paid plans, and Copilot Free is left out.

Learn more about GitHub Copilot

  1. 4Vibe coding: a free course for beginners · Lesson 4 of 4GitHub Copilot agent mode: build in VS Code, then hand work to the cloudUse GitHub Copilot agent mode in VS Code to build a small app from one prompt, review each edit, then hand an issue to the cloud agent for a pull request.10 min
  2. 1Vibe coding: a free course for beginners · Lesson 1 of 4What is vibe coding? Describe it, run it, check itVibe coding is building software by describing it to an AI and judging what runs, not the code. The loop, the term's origin, the risks and the tools.9 min

Frequently asked questions

Is GitHub Copilot free?

Partly. The Free plan has no charge, with limited features and some AI credits each month. Verified students get Copilot Student at no cost, and the paid plans bring a wider choice of models and more credits.

Does GitHub Copilot work outside VS Code?

Yes. Besides VS Code it works in Visual Studio and the JetBrains editors, and it also runs in a terminal, in a desktop app, and on GitHub.com and GitHub Mobile. The steps above use VS Code because it is easy to start with.

What is the difference between agent mode and the Copilot cloud agent?

Where they run. Agent mode works in your editor on your computer while you watch and approve. The cloud agent works on GitHub in the background and returns a pull request for you to review later.

Which AI models does GitHub Copilot use?

Several. It began on OpenAI Codex in 2021, added a choice of models in 2024 and now offers ones from OpenAI, Anthropic, xAI and Google.

Should I use GitHub Copilot or Cursor?

It comes down to whether you will switch editors. Copilot installs into VS Code or JetBrains, while Cursor is an app of its own. Both let an agent edit files and run commands.

Can Copilot change my files without asking?

Not on the default setting. Manual permissions asks before any action that no approval setting already covers. The two looser levels skip confirmation, so keep them for a workspace you trust.

Sources

  1. GitHub Docs: About GitHub Copilot (read November 6, 2026)https://docs.github.com/en/copilot/get-started/what-is-github-copilot
  2. GitHub Docs: Plans for GitHub Copilot (read November 6, 2026)https://docs.github.com/en/copilot/get-started/plans
  3. GitHub Docs: Quickstart for using GitHub Copilot on GitHub.com (read November 6, 2026)https://docs.github.com/en/copilot/get-started/quickstart
  4. GitHub Docs: GitHub Copilot surfaces (read November 6, 2026)https://docs.github.com/en/copilot/concepts/copilot-surfaces
  5. Wikipedia: GitHub Copilot (read November 6, 2026)https://en.wikipedia.org/wiki/GitHub_Copilot
  6. VS Code docs: Set up GitHub Copilot in VS Code (read November 5, 2026)https://code.visualstudio.com/docs/setup/copilot
  7. VS Code docs: Quickstart, complete your first task with an agent (read November 5, 2026)https://code.visualstudio.com/docs/agents/quickstart
  8. VS Code docs: Review and revert agent changes (read November 5, 2026)https://code.visualstudio.com/docs/agents/run/review-code-edits
  9. VS Code docs: Manage approvals and permissions (read November 5, 2026)https://code.visualstudio.com/docs/agents/run/approvals
  10. GitHub Docs: Using Copilot in the cloud on GitHub (read November 5, 2026)https://docs.github.com/en/copilot/how-tos/use-copilot-agents/cloud-agent/use-cloud-agent-on-github
  11. GitHub on YouTube: How to use GitHub Copilot (the complete beginner's guide) (read November 6, 2026)https://www.youtube.com/watch?v=SJqGYwRq0uc
  12. Powercademy on YouTube: How To Use GitHub Copilot in 2026 (read November 6, 2026)https://www.youtube.com/watch?v=5Lg_bG907k8

Share this lesson

Build with agent mode

The Copilot lesson goes from sign-in to a reviewed edit, permission levels and the cloud agent.

Start the lesson

This lesson as Markdown · All tools · All courses