Atlas is taking you through this one. Free, and it runs right here. No signup.
Learn Cursor in 5 minutes.
Cursor is a fork of VS Code with a model wired into the editor. The important part is how to use AI within Cursor: Tab, Cmd+K, and Cmd+I, and the difference between the three.
Lesson not loading? Open it in a new tab. It is the same free lesson either way.
Tab, Cmd-K, or Agent?
These are not interchangeable, and picking the wrong one is the usual reason people try Cursor once and go back to their old editor.
Predictive edits
Autocomplete that can jump. It predicts your next edit, including somewhere else in the file, and pressing Tab again hops there. Best for changes you already know how to make and just want typed faster.
Inline edit
Select some code, describe the change in plain English, get a diff to accept or reject. Scoped to the selection, so it's the safe choice when you know exactly which lines should change.
Agent
Give it a goal and it works across the repo, reading files, editing in several places, running commands. The right tool when a change spans files you have not opened yet.
What you will do in the lesson
You answer and try things. Nobody demonstrates at you for five minutes.
- Tab, and when to ignore it. Accepting every suggestion is how you end up with code you don't understand. You'll practise reading a suggestion before taking it.
- Cmd-K on a real selection. Rewriting a function from an instruction, then actually reading the diff. This is the habit that keeps you in control of the change.
- Giving the Agent a bounded task. Why “rename this concept everywhere” works well and “make the app better” doesn't.
- Context with @-mentions. Pointing at the specific files, docs or symbols that matter instead of hoping the model guesses.
- Knowing when to stop. Spotting the point where the model is guessing and you should take the keyboard back.
Before you start
Is this just VS Code with autocomplete?
It's a VS Code fork, so extensions, keybindings and themes carry over and the editor feels familiar on day one. The difference is scope: the model can read and edit across files rather than only finishing the line under your cursor.
Do I need to install anything for this lesson?
No. The lesson runs in the frame on this page, with no signup and no download. Installing Cursor afterwards is optional, and the basic tier is free.
Does Cursor send my code somewhere?
By default, requests go to Cursor's servers for inference. Privacy Mode prevents your code being retained, and is worth switching on before you point Cursor at a work repository. Check your employer's policy first.
I already use GitHub Copilot. Is this different?
Copilot is strongest as an inline assistant inside an editor you already have. Cursor bets on the agent loop across a whole repository. They overlap most on completion and diverge most on multi-file work. Trying both is the fastest way to tell which one fits how you work.
Other tools people learn next
Same format, same five minutes, still free.
Learners from these organizations are already learning on 
