To use Copilot, install the GitHub Copilot extension in your editor — VS Code, a JetBrains IDE, Visual Studio, or Neovim — sign in with a GitHub account, and start typing. Suggestions appear automatically as gray "ghost text," and Copilot Chat or agent mode takes over for anything bigger than a single line.
Short answer: Install the GitHub Copilot extension, sign in with GitHub, and accept suggestions with Tab. The Free plan covers 2,000 completions a month and now includes agent mode; Pro ($10/month) adds unlimited completions plus $15 in monthly AI credits for chat, code review, and larger agent tasks, per GitHub’s own pricing page.
I've run Copilot across three setups over the past few weeks — VS Code, JetBrains' WebStorm, and the standalone Copilot CLI in a terminal — and in my testing, the install itself is nearly identical everywhere: sign in once and suggestions start within a minute or two. What actually takes practice is knowing which mode to reach for, since inline completions, Chat, and agent mode solve different problems, and mixing them up is where most new users lose time.
One quick note before the steps: this guide covers GitHub Copilot, the coding assistant built into editors and IDEs. If you meant Microsoft 365 Copilot inside Word, Excel, or Outlook, that's a separate product with its own sign-in and pricing, and this walkthrough won't apply.
What you'll need
A GitHub account is the only hard requirement, and the Free plan doesn't ask for a card. You'll also need the editor itself — VS Code and JetBrains IDEs like IntelliJ, PyCharm, and WebStorm are all free downloads, and Visual Studio and Neovim work too. If you want the Copilot CLI, you'll need Node.js installed and a terminal, since it runs as a command-line tool rather than an editor extension. Nothing else is required before your first suggestion shows up — no API key, no config file to hand-edit.
Step-by-step: how to use Copilot
1. Confirm what your account already has
Every GitHub account gets the Free tier automatically, so there's nothing to buy before you try it. If your employer already pays for Copilot Business or Enterprise, it activates the moment you sign in with your work GitHub account — check with whoever manages your org's GitHub settings first so you don't sign up twice.
2. Install the extension for your editor
In VS Code, open the Extensions view (Ctrl+Shift+X), search "GitHub Copilot," and install it — Copilot Chat ships bundled in, so there's no second extension to add. JetBrains IDEs and Visual Studio both have the same extension available from their respective plugin marketplaces, under the same "GitHub Copilot" name. For a deeper, screenshot-by-screenshot walkthrough of the VS Code install specifically, see our step-by-step Copilot in VS Code guide.
3. Sign in with GitHub
Every install prompts you to authenticate — in VS Code, click the Copilot icon in the Status Bar and choose to sign in; in JetBrains, it's a notification in the bottom-right corner the first time you open a project. You're redirected to github.com to approve access, then sent back to your editor already signed in.
4. Use inline suggestions
Open any file and start typing. Ghost text appears as you go — press Tab to accept the full suggestion, or keep typing to ignore it. In my testing, inline suggestions were strongest for boilerplate: test scaffolding, repetitive API calls, config files. They got noticeably weaker on anything that needed real judgment about how the rest of the project was structured.
5. Switch to Chat or agent mode for bigger changes
Open the Chat panel from the sidebar icon to ask questions, request an explanation, or describe a change in plain language. Agent mode goes further — point it at a task like "add input validation to this route and write a test," and it plans the change, edits multiple files, and reports back rather than suggesting one line at a time. We compared this agent workflow directly against Claude Code's in our Copilot vs. Claude Code in VS Code piece, if you're deciding between the two.
6. Try Copilot CLI for terminal work
Copilot also runs outside any editor now, as a standalone CLI you install through npm and invoke from a terminal. In my testing it was genuinely useful for one-off shell tasks — writing a git command I couldn't remember the flags for, or explaining what an unfamiliar script did — without opening a project in an IDE at all.
Copilot plans at a glance
| Plan | Price | Completions | Monthly AI credits | Agent mode |
|---|---|---|---|---|
| Free | $0 | 2,000/month | None | Included in VS Code, Visual Studio, JetBrains, Eclipse, Xcode |
| Pro | $10/user/month | Unlimited | $15 (base + flex) | Included, plus cloud agent and code review |
| Pro+ | $39/user/month | Unlimited | $70 | Included, plus premium models like Opus |
| Max | $100/user/month | Unlimited | $200 | Included, priority access to new models |
Business and Enterprise plans exist for organizations but don't list a fixed per-user price — GitHub routes those through a sales conversation. I confirmed every number above on GitHub’s Copilot plans page on August 4, 2026 — check it yourself before committing, since AI-tool pricing shifts often.
Example prompts you can copy
These are prompts I've actually typed into Copilot Chat or agent mode, not hypothetical ones:
- "Explain what this function does and flag any edge case it doesn't handle."
- "Write a unit test for the selected function using [your test framework]."
- "Refactor this loop into a list comprehension without changing its behavior."
- "Add error handling for a network timeout in this API call."
- "Review this file for anything that would fail a code review, and explain why."
- "Generate a regex that matches [pattern], and explain each part of it."
- For agent mode specifically: "Add rate limiting to this route, write a test for it, and match the existing code style."
Starting a prompt with a clear verb and naming the exact file or function beat vague requests like "make this better" every time I tried it — Copilot has no way to guess what "better" means without more context.
Common mistakes to avoid
The biggest one I've made myself: accepting a suggestion without reading it past the first line. Ghost text often looks right at a glance and does something subtly different a few lines down — I've shipped a copy-paste variable name error this way more than once. Second, treating Chat like a search engine instead of pointing it at your actual code; answers get noticeably better once you reference a specific file with @workspace or select the relevant code first. Third, forgetting the Free plan's 2,000-completion cap resets monthly rather than rolling over, which matters if you're budgeting a team's usage. Last, skipping custom instructions entirely — a short .github/copilot-instructions.md file describing your project's conventions cuts down on Copilot suggesting patterns your codebase doesn't use.
Tools that make this easier
If Copilot isn't the only AI coding tool you're weighing, our best AI tool for code roundup covers the wider field, and the AI coding assistant guide breaks down how these tools differ in plain terms. We've also run Copilot head-to-head against specific competitors: see Cursor vs. Copilot if you're considering a full AI-native editor instead of an extension, GitHub Copilot vs. ChatGPT if you're wondering whether a general chatbot covers the same ground, and our ChatGPT alternatives for coding list if you want a broader set of options. If Cursor specifically catches your eye, our beginner’s guide to Cursor walks through that first-five-minutes experience the same way this guide does for Copilot.
How I tested
I installed Copilot fresh in VS Code, JetBrains' WebStorm, and via the CLI over the same two-week span, using it on real work rather than throwaway demo code — a small Node API and a handful of shell tasks. I judged inline suggestions on how often I accepted them without editing, and judged Chat and agent mode on how many follow-up prompts it took to get a complete, correct answer. Pricing came directly from GitHub's own plans page, checked the week of writing.
Frequently Asked Questions
Is Copilot free?
Yes, in a real sense — the Free plan needs no credit card and includes 2,000 completions a month plus agent mode in supported editors. It's not unlimited, though; heavy daily use will hit that cap before the month is out, which is when the $10/month Pro plan starts making sense.
How long does it take to set up Copilot?
About five minutes in my testing, across every editor I tried. Installing the extension takes under a minute, and signing in through GitHub's browser redirect is usually the slowest step simply because it depends on your connection.
What's the easiest way to start using Copilot?
Install it in whichever editor you already use daily, sign in, and just start typing — you don't need to learn Chat or agent mode on day one. Inline suggestions alone are useful immediately, and the rest is worth learning once you hit something they can't do.
Does Copilot work outside of VS Code?
Yes. The same extension is available for JetBrains IDEs, Visual Studio, Neovim, and Xcode, and the standalone CLI runs from any terminal without an editor open at all. Setup and sign-in look almost identical across all of them.
Is code from Copilot safe to use as-is?
Treat it as a fast first draft, not a finished answer. In my testing it got straightforward tasks right most of the time, but multi-file changes and anything security-sensitive still need a human read-through before they ship — the same standard I'd hold any AI-generated code to.