Anysphere · Free course

Learn Cursor.

A free course on working well in Cursor, from Agent modes and Tab to rules, skills, subagents and Cloud Agents.

6 modules11 lessons38 techniques
The Cursor editor, with an agent chat beside open code.
Syllabus

What's in this course.

Go in order, or jump to the lesson you need today. Each technique has a prompt you can copy.

  1. 01

    Basics

    Move in from VS Code, pick an Agent mode, and learn Tab, Cmd+K and how to undo what Agent changed.

  2. 02

    Coding

    Generate code that fits your repo, debug with Debug mode and the built-in browser, and refactor in steps you can review.

  3. 03

    Workflows

    Steer Agent with a plan and checkpoints, redirect it mid-run, set what it may run alone, and review before you commit.

  4. 04

    Hidden Features

    Learn rules, skills, subagents, Cloud Agents and the way Agent searches your code.

  5. 05

    Advanced

    Pick a model and context size, use @ mentions, read the context ring, and ask side questions without derailing Agent.

  6. 06

    Mistakes

    Find out why Agent edits the wrong file, duplicates your code or goes in circles, and how to fix each one.

Check yourself

How well do you know Cursor?

Eight real situations, from your first task to running several agents at once. You get the reason behind each answer as soon as you choose. No sign-up.

Take the quiz
How to use Cursor well

Agent is now the center of Cursor

Cursor is built on the VS Code codebase, so your settings, themes and keybindings import in one step. Extensions come from a different store, so check the ones you depend on. Press Cmd+I (Ctrl+I on Windows and Linux) to open Agent, and try it on a repository you know, because then you can tell when it's wrong.

If you're following an older guide, note that inline edit is gone. Select code and press Cmd+K or Cmd+L (Ctrl on Windows and Linux), and Agent opens with those lines referenced in its input. Tab is still the autocomplete, and it can edit several lines, add imports and jump to your next edit. Press Shift+Tab in the chat input to rotate through the modes: Agent for most work, Ask for answers with no edits, Plan to approve an approach, and Debug for bugs you can't explain.

Tell Agent where to look

Agent finds code with its own search tools, so name an exact symbol when you know one and describe the behavior when you don't. Use @ to attach files, @Terminals output or a Git diff when you already know what's relevant. Point to one existing file as the pattern to copy, because Agent follows your patterns far more reliably when you say which one.

Always-on rules, skill descriptions and MCP catalogs take up context before you type anything. The context ring next to the prompt input shows how full the window is, and clicking it shows what's using the room. Start a new chat when you switch tasks or when Agent keeps going in circles.

Rules for what it should know, skills for what it can do

Add a rule when Agent repeats a mistake, not before. Keep it short and concrete: real commands, your conventions, and what it must not touch. Project rules live in .cursor/rules, and AGENTS.md works for simple instructions in plain markdown. Cursor's course draws the line this way: rules are for what the agent should always know, and skills are for what it can do when asked. Anything needed only sometimes belongs in a skill, which you run by typing / and its name.

Run agents and keep control

For a change across several files, use Plan mode, edit the plan, and click Build. If the result misses, revert to a checkpoint and make the plan more specific, which Cursor's docs say is often faster than follow-up prompts. Choose how much runs without asking under Settings > **Agents > Approvals & Execution, where Cursor's docs recommend Auto-review** for most people.

Background Agents are now called Cloud Agents. Each runs on its own virtual machine, so you can close your laptop and review a pull request later. Before you commit local work, run /agent-review, and use /worktree to try a risky change without touching your current checkout.

Frequently asked
Is Cursor free?

Cursor has a Free plan. New customers buy paid plans through The X Subscription: Plus at $8, Premium at $30, Super at $100 and Ultra at $200 a month, as listed in November 2026. Included usage covers Cursor's own models, Grok and Composer. Claude, GPT and Gemini are paid from prepaid credit on Premium or higher.

What does Cmd+K do in Cursor now?

It sends your selected code to Agent. Cursor's help says Cmd+K used to open inline edit, which has been replaced by this workflow. Select code, press Cmd+K or Cmd+L (Ctrl+K or Ctrl+L on Windows and Linux), type the change in Agent's input, and it applies as a normal Agent edit you can review.

What happened to Cursor Background Agents?

They're now called Cloud Agents. To start one, select Cloud in the dropdown under the Agent input, go to cursor.com/agents, or mention @cursor in Slack, GitHub or Linear. Each runs on its own virtual machine, so it keeps going after you close your laptop, and you review a pull request later.

What is the difference between Agent, Ask, Plan and Debug in Cursor?

Agent is for most work and edits your code. Ask gives answers with no edits. Plan lets you approve an approach before a multi-file change. Debug is for bugs you can't explain, and it adds logs and reads what actually happened before changing code. Press Shift+Tab in the chat input to rotate through them.

What are Cursor rules, and how are they different from skills?

Rules are standing instructions for what the agent should always know, such as your commands and conventions, and project rules live in .cursor/rules. Skills are for what it can do when asked. A skill is a folder containing a SKILL.md file, and you run it by typing / and its name, or let Agent load it when relevant.

Can I use my VS Code extensions and settings in Cursor?

Mostly, yes. Open Cursor Settings with Cmd+Shift+J (Ctrl+Shift+J on Windows and Linux), go to General, then Account, and click Import under VS Code Import. That brings over extensions, themes, settings and keybindings. Extensions come from a different store, so check the ones you depend on.

Why does Cursor edit the wrong file?

Agent takes your request literally, and where you give no intent, it uses its best judgment. A wrong edit usually means it had to guess the file, the pattern or the limit. Name the target with an @ mention, point to one example to follow, and say what it must not touch.