What is Cursor AI? What it does, who it is for and what it costs

What is Cursor AI?

Cursor describes itself on its homepage as your coding agent for building ambitious software. In plain terms it is a code editor with an AI agent built in. You open a project folder, describe what you want in a chat panel, and the agent searches your files, makes the edits and shows you a diff to accept or reject. It can also run terminal commands such as your tests.

It is made by Anysphere. On August 14, 2026, Cursor's blog announced that Cursor is now part of SpaceX, completing an acquisition that began with a partnership in April. The product, docs and download are still at cursor.com. Everything below comes from Cursor's own docs and pricing pages, read on October 30, 2026.

Cursor is not a website builder. It does not host your app or give you a database. It works on files on your computer, which means the code is yours to keep, host anywhere and open in other tools. That is the main difference from a browser builder such as Lovable or Bolt.

What does Cursor AI actually do?

Five features cover most of what people use it for.

1. Tab. Tab is autocomplete. It suggests code as you type based on your recent edits, surrounding code and linter errors. You press Tab to accept and Escape to reject. It can edit several lines, add missing imports and jump to your next edit location, including in other files.
2. Agent. Agent is the assistant that completes tasks on its own: it searches files, edits code, runs terminal commands and fixes errors. Open it with Cmd+I on Mac or Ctrl+I on Windows and Linux. Its docs say there is no limit on the number of tool calls it can make in one task.
3. Plan Mode. Press Shift+Tab in the agent input and Cursor researches your codebase, asks clarifying questions and writes a plan you can edit before any code changes. Cursor's Plan Mode docs recommend it for features that touch many files or have unclear requirements.
4. Context with @. Type @ in the chat to attach files, folders, terminal output, earlier chats, git diffs or the built-in browser, so the agent looks at the right things.
5. Rules. Rules are saved instructions the agent follows in your project, such as your coding conventions. The Cursor rules guide shows the format and examples.

Around those sit extensions that matter once you are past the basics: skills, subagents, hooks and MCP servers for connecting outside tools, all listed under Customize in the docs. Agent changes can be undone: hover a previous message and click Restore Checkpoint to roll back everything the agent changed after that point. For the agent itself, see the Cursor Agent guide.

Who is Cursor for, and who should skip it?

If your goal is a working app from a description with no setup, compare it in Claude Code vs Cursor vs Lovable. If you are choosing between editors, read Cursor vs Windsurf or Cursor vs Copilot.

How is Cursor different from Claude Code and from app builders?

Three working styles get compared constantly, and Cursor sits in the middle. The table uses only what each vendor says about itself.

That makes Cursor a good fit when you want to keep the code and also see it. For a deeper comparison with a terminal agent and the builders, read Claude Code vs Cursor vs Lovable, and see Cursor alternatives if you want to shop around.

A few limits are worth knowing before you commit. Usage is capped by plan, and third-party models are billed on top of the monthly price. The extension registry is Open VSX, so a specific VS Code extension may not be there. And an agent can be confidently wrong, which is why Cursor's own quickstart ends with reviewing the diff and running your project's checks. None of these is a reason to skip it. They are reasons to start on the free plan with a small task.

Which AI models does Cursor use, and what does it cost?

Cursor's homepage says you can choose between models from OpenAI, Anthropic, Gemini, SpaceXAI and Cursor itself, and you can switch models at any point in a chat. Its pricing page lists these individual plans as of October 2026:

The page adds that third-party model usage in Cursor is not included in the monthly prices and is billed separately by usage, and a Teams plan starts at USD 30 per user. Prices exclude taxes. Plans have changed several times, so check the live page before paying. The Cursor pricing guide breaks the tiers down in more detail.

Is your code safe in Cursor?

Cursor's privacy page says that when you use AI features it sends prompts and code context to model providers such as OpenAI, Anthropic and Google. With Privacy Mode enabled, your code is never used for training. You turn it on in Cursor Settings (Cmd+Shift+J on Mac, Ctrl+Shift+J on Windows and Linux) under General. For teams it is on by default.

• Add a .cursorignore file in your project root to keep files out of AI context. Cursor already ignores .env files, .git/ and lock files by default.
• Ignored files are blocked from Agent, but terminal commands and MCP tools run outside Cursor's file access controls and may still read them, so keep real secrets out of the project folder.
• Review each diff. An agent can make a confident change that is wrong.

How do you try Cursor in ten minutes?

Download Cursor, sign in, open any folder and press Cmd+I or Ctrl+I. Start by asking it to explain the code, then make one small change. If you have no project, take an idea from the ideas board: each has a generated build prompt, so you can paste it into an empty folder and watch Cursor scaffold the first version. Step-by-step instructions are in how to use Cursor.

I am starting a new project in this empty folder.

Build a small web app for a tutor who tracks which students have paid for the month. One list page with student name, amount and a paid/unpaid toggle, plus an add-student form. Store data locally for now. No accounts.

Before writing code, tell me which stack you will use and list the files you will create. Wait for my go-ahead. After that, build it and tell me the exact command to run it.