Set up Copilot in VS Code

Installing Copilot in VS Code is a few clicks. This page covers those clicks, then sets one piece of framing that shapes how you think about the tool for the rest of the course: what you installed is not a single feature but one assistant that shows up on several surfaces.

Getting Copilot into VS Code

On a current VS Code build, the Copilot experience is built in: the AI features ship with the editor itself (the setup documentation treats removing them as a setting, not an uninstall). If your build predates that or arrived without them, the marketplace path applies: open the Extensions view, search for GitHub Copilot, and follow the prompts your version shows. Either way, do not treat a companion extension (a separate Copilot add-on installed alongside the editor) as a permanent requirement - the packaging has shifted more than once.

Then you sign in. Copilot is tied to a GitHub account and a Copilot plan. Sign-in starts from the Copilot icon in the Status Bar (a first run may prompt you itself) and completes in the browser - choose the sign-in method your account offers, which on enterprise setups may differ from a plain GitHub login. If your employer provides Copilot, access comes through your organization’s plan: you sign in with the licensed account, and the policies that come down with it - which features are on - partly decide what you can do. That is worth noticing now, because at a locked-down shop what Copilot can do is partly decided above your account. The specific places that bites, like organization rules and model availability, come up where they matter in later chapters.

Because the packaging and the sign-in flow shift over time, treat the official VS Code Copilot setup docs as the source of truth for the current setup on your machine (verified September 2026).

One assistant, several surfaces

Here is the framing. The thing you installed is one assistant that reaches you through more than one surface in the editor, and the surfaces feel different enough that people mistake them for separate tools:

• Completions - the grey ghost text that appears inline as you type, suggesting the rest of a line or block. The original Copilot experience, still present as of September 2026.
• Chat - a conversational panel where you ask questions, request changes, and hand off whole tasks. This is where the agent (the autonomous Copilot this course is about) lives, and where this course spends its time.
• Others, later - inline chat in the editor, and the cloud coding agent that works on GitHub itself without your editor open at all.

It is all the same Copilot reading the same project, shaped by the same instruction files you will write in the Rules chapter. The surface is just where you happen to be standing - so the instinct “I learned the chat panel, now I have to learn this other thing” is mostly wrong. What you learn about steering Copilot carries from one surface to the next.

Confirm it’s alive

Two quick checks before moving on. Open a code file in the repo you are working in - the course example repo (introduced on the chapter page, orders-service), or your own - and start typing a line; ghost text should offer a completion within a second or two. Then open the Chat view (its icon in the Activity Bar, or from the Command Palette) and ask something harmless about the project: “what does this service do?” A real answer means it is installed, authorized, and reading your workspace. If a check comes up empty, suspect sign-in or an organization policy turning that feature off rather than a failed install; the setup docs linked above cover the first, and your admin can tell you which features your plan enables.

Copilot is running and can see your code. Next, the distinction that trips people up in their first week: those two surfaces - completions and chat - are not interchangeable, and reaching for the wrong one is the most common early mistake. Next: completions vs chat.