To use GitHub Copilot in VSCode, install the GitHub Copilot extension from the Extensions view, sign in with your GitHub account through the Status Bar icon, and start typing — inline suggestions appear automatically, and Copilot Chat handles anything that needs more than one line of code. Free access covers casual use; the $10/month Pro plan removes the completion cap and adds agent mode for multi-file changes.
Short answer: Install the GitHub Copilot extension in VS Code, click the Copilot icon in the Status Bar, and sign in with GitHub. Free gives you 2,000 completions a month; Pro ($10/month) adds unlimited completions plus chat, agent mode, and code review. Accept suggestions with Tab, or open Copilot Chat for anything bigger than one line.
I've had Copilot running in VS Code for months now, across a Flask project and a couple of front-end repos. In my testing, the setup itself takes about five minutes — the part that actually takes practice is learning which chat mode to reach for. Here's the real walkthrough, what each plan costs, and the mistakes that waste your first week with it.
What you'll need
A GitHub account is the only hard requirement — Copilot's Free plan works with any account and doesn't ask for a card. You'll also need VS Code itself, which is free from Microsoft, and a project folder open so Copilot has something to read. If you want unlimited completions, chat, and agent mode without the Free plan's monthly caps, budget $10/month for Pro. Nothing else is required: no separate API key, no config file to hand-edit before your first suggestion shows up.
Step-by-step: how to use GitHub Copilot in VSCode
1. Confirm your Copilot access
Every GitHub account gets the Free tier automatically, so there's nothing to sign up for before you open VS Code. Free includes 2,000 code completions a month plus limited chat access, according to GitHub’s Copilot plans page. If your organization already pays for Pro, Pro+, Business, or Enterprise, Copilot activates the moment you sign in.
2. Install the extension
Open VS Code, go to the Extensions view (the square icon in the sidebar, or Ctrl+Shift+X), search "GitHub Copilot," and install it. Copilot Chat now ships bundled with the same extension, so there's no second install needed to get chat working.
3. Sign in
Hover over the Copilot icon in the Status Bar at the bottom of the window and choose Use AI Features, then sign in with your GitHub credentials when prompted. If you don't already have a Copilot subscription, VS Code enrolls you in the Free plan automatically, per Visual Studio Code’s own setup docs.
4. Try inline suggestions
Open any file and start typing — gray "ghost text" suggestions appear as you go. Press Tab to accept a full suggestion, or Ctrl+Right (Cmd+Right on Mac) to accept just the next word or line if you only want part of it. In my testing, inline suggestions were strongest for boilerplate — test scaffolding, repetitive API calls, config files — and weakest on anything that needed real judgment about the project's architecture.
5. Open Copilot Chat
Click the Chat icon in the sidebar, or press Ctrl+Alt+I, to open the chat panel. This is where you ask questions about your code, request explanations, or describe a change in plain language instead of writing it yourself.
6. Pick a chat mode
Copilot Chat runs in three modes: Ask for questions and explanations without touching your files, Edit for targeted changes to files you point it at, and Agent for open-ended tasks where Copilot plans the approach, edits across your project, and runs commands on its own. Start with Ask when you're learning a new codebase, and save Agent for tasks you're comfortable reviewing afterward — it moves fast and doesn't check in before every file it touches.
7. Set up project context (optional but worth it)
Type /init in a chat session and Copilot scans your codebase to write a custom instructions file, so future answers already reflect your project's conventions instead of guessing. I do this once per repo, and it noticeably cuts down on Copilot suggesting patterns that don't match the rest of the codebase.
Example prompts you can copy
Specific, scoped prompts get better results than vague ones in every chat mode. A few that work well as a starting point:
- "Explain what this function does and flag anything that looks like a bug."
- "Add error handling to this API route without changing its signature."
- "Write a unit test for
utils/parse.pyusing the same test framework already in this repo." - "Refactor this component to remove duplicate state, but don't rename any props."
- "Review this diff for security issues before I open a pull request."
Notice what each one shares: a specific file or function, a clear finish line, and something explicitly off-limits. That last part matters more than it looks — telling Copilot what to leave alone cuts down on surprise edits, especially in Agent mode.
Common mistakes to avoid
The most common mistake is leaving Copilot in Agent mode for a small, well-defined fix — Ask or Edit mode gets there faster and touches less. Second, accepting a long inline suggestion without reading it; Copilot is good at pattern-matching your existing code, not at knowing your business logic, and a smoothly-formatted suggestion can still be wrong. Third, skipping the /init step on a large codebase and then wondering why chat answers feel generic — a few minutes of setup fixes that. Fourth, staying on the Free plan's 2,000-completion cap without realizing it, then being confused when suggestions stop mid-month; Pro removes that ceiling for $10/month. Fifth, treating every chat mode as interchangeable — Agent is the most capable but also the one most likely to make a change you didn't expect, so I keep it for tasks I'm ready to review, not quick one-line fixes.
GitHub Copilot plans in VSCode
| Plan | Price | What you get |
|---|---|---|
| Free | $0/mo | 2,000 completions/month, limited chat, Copilot CLI |
| Pro | $10/user/mo | Unlimited completions, full chat and agent mode, $15/mo in model credits |
| Pro+ | $39/user/mo | Premium models including Opus, $70/mo in credits, audit logs |
| Max | $100/user/mo | Priority access to new models, $200/mo in credits, built for heavy agent use |
| Business / Enterprise | Contact sales | Org-wide policy controls, centralized billing, SSO |
I confirmed these directly on GitHub’s Copilot plans page on August 3, 2026 — plan names and credit allowances in this category shift often, so check the current page before you commit. For most individual developers, Pro at $10/month is the plan that actually removes the friction: unlimited completions plus real chat and agent access, without jumping straight to Pro+'s premium-model pricing.
Tools that make this easier
Copilot isn't the only AI coding tool worth having open, and it isn't always the strongest one for heavy multi-file work. My tested Cursor vs Copilot comparison ran the same coding brief through both tools' agent modes and found Cursor a step ahead on large refactors, with Copilot winning on price and staying inside the editor you already use. For a wider view of what's out there, best AI tool for code covers the field with real prices, and AI coding assistant is a good primer if you're new to this category altogether. If you want to look beyond GitHub's own stack, ChatGPT alternatives for coding rounds up the competition. Coming from OpenAI's side of things, how to use ChatGPT Codex walks through a close cousin of Copilot's agent mode, built into ChatGPT instead of VS Code. And if a full AI-native editor appeals to you more than an extension, how to use Cursor for beginners covers getting started there.
My take
Copilot's real strength is that it asks nothing of your workflow — install the extension, sign in, and it's already working inside the editor you use every day. The Free plan is genuinely usable for light work, and Pro at $10/month is cheap enough that most developers should just try it rather than debate it. Where I'd pump the brakes is Agent mode on anything you haven't scoped clearly; it's capable, and capable-but-unsupervised is exactly the combination that produces changes you didn't ask for.
Frequently Asked Questions
Is GitHub Copilot free to use in VSCode?
Yes — the Free plan gives you 2,000 code completions a month and limited chat access with no card required. Unlimited completions, full agent mode, and code review start on the $10/month Pro plan.
How long does it take to set up GitHub Copilot in VSCode?
About five minutes: install the extension, sign in through the Status Bar, and suggestions start appearing as soon as you open a file. Learning which chat mode fits which task takes longer — plan on a week of regular use before it feels automatic.
What is the easiest way to use GitHub Copilot in VSCode well?
Start with Ask mode for questions and Edit mode for small, targeted changes, and save Agent mode for tasks you're ready to review carefully. Write prompts that name a file or scope and mark one thing not to touch — it cuts down on surprise edits.
Does GitHub Copilot in VSCode work with languages other than Python and JavaScript?
Yes. Copilot supports most mainstream languages — Python, JavaScript, TypeScript, Go, Java, C#, Ruby, and more — since it works from the code and context already in your project rather than a fixed language list.
Can I use GitHub Copilot in VSCode without paying?
Yes, the Free plan works with any GitHub account and includes 2,000 completions a month plus limited chat. It's enough to try Copilot properly before deciding whether the $10/month Pro plan is worth it for your workload.