Cursor overview

Cursor presents a major shift in the software development landscape. Instead of being a text editor with smart AI plugins, it's a complete IDE built for AI-powered development. It integrates AI into its core for a seamless experience.

This topic covers the first steps with Cursor. We'll explore how to install, navigate, and configure it for AI-driven development. You'll learn about its "AI-first" design and how it understands your work, helping you solve problems, debug faster, and write cleaner code.

Getting started with Cursor

Cursor is widely known as an AI-first IDE. Traditional code editors usually add AI capabilities through optional plugins or extensions. Cursor is different because it weaves these capabilities into its core from the beginning. It is a fork of VS Code, which makes it feel immediately familiar to most developers. However, Cursor has been fundamentally altered to prioritize and streamline the flow of information to and from an underlying LLM.

This deep, core integration defines the "AI-first" experience. It means LLMs are not just external helpers; they're a seamless part of the IDE's internal state. They automatically have direct, live access to crucial elements of your workspace. This includes your currently active code selection, the entire project file structure, and even recent terminal output or error messages.

This rich context allows the underlying LLMs to formulate answers based on the current state of the application. This eliminates the friction of manually describing the environment, as the LLM operates with the same visibility as the developer.

Because it is based on VS Code, it supports the same ecosystem of extensions. You don't have to give up your favorite themes or language tools. You get the familiarity of a standard editor combined with an AI-powered partner. This hybrid approach makes the transition seamless for developers.

Installing Cursor

Cursor is a standalone application, meaning you install it just like any other program on your computer. It is available across major operating systems, including Windows, macOS, and Linux. This standalone approach gives the editor's indexing and context engines access levels that standard plugins don't have.

To begin, navigate to the official website and download the correct installer package for your operating system. Running the installer and following the prompts is all you need to install it:

During the initial setup, the application provides a migration utility to help you import settings, keybindings, and extensions from VS Code. You can always import later via Cursor Settings > General > Import Settings from VS Code:

Once you complete the installation, the application guides you through the necessary authentication process. When you install Cursor for the first time, you're automatically in the default free plan. Your Pro one-week trial begins immediately, allowing you to try all the features. After that, you can upgrade to a paid tier.

As a final step, you can set up Cursor's command-line tools. This allows you to launch Cursor from the CLI using cursor or code. You can use it to open directories or files directly from your terminal, just as you would with other standard editors:

You can install the CLI commands via the command palette, which we will see later.

Exploring the interface

When you first open Cursor, you will notice a layout that looks familiar if you have used VS Code, but with a few key modifications. The interface is designed around two primary views that you can toggle between using ⌘/Ctrl E:

The first is the Agent layout. This is often the default experience when you start a new project or begin a task. In this view, the traditional Editor pane—where you write and edit code—is typically minimized or replaced by a dedicated workspace for the Agent. This workspace focuses on the chat panel. It gives you a consolidated view of the Agent's thought process, planned changes, and execution output.

Then there is the standard Editor view. This is your default coding environment, similar to any traditional IDE. You will see a large, central editor pane where you open and modify your code files directly. It is the place for deep focus when you want to write or manually review code yourself.

The Panel is typically positioned at the bottom of the screen, summoned with ⌘/Ctrl J. It is where you find the standard Terminal for executing operating system commands, the Output for logs, and the Debug Console:

The Status Bar, which stretches across the very bottom edge of the IDE, constantly provides real-time feedback. It displays important contextual information such as the current Git branch, the selected programming language, line and column numbers, and various notifications from the IDE.

Core Cursor configuration

The next crucial step is customizing the environment to match your specific workflow and standards. Like VS Code, Cursor uses both user and workspace settings. User settings apply globally to all your projects, covering things like themes and general editor behavior. Workspace settings are saved inside a .vscode folder within your project, making them perfect for defining project-specific rules that you can commit to version control and share with your team.

To quickly access any setting or command, the Command Palette is your primary navigation tool. You can open it using the shortcut ⌘/Ctrl Shift P. The palette is a search bar where you can type in commands like "Keyboard Shortcuts", "Install 'cursor' to shell", or "Cursor Settings." It removes the need to remember where a specific feature is located in the nested menus:

A top priority is managing the underlying LLM. You can manage models available for your project under Cursor Settings > Models. For maximum flexibility, you can also bring your own key (BYOK) for direct control over your API costs and usage limits. This allows you to integrate different models and model versions from providers like OpenAI or Anthropic, using your own API key. Once set, you can select from a variety of supported language models via a dropdown in the chat interface:

It is important to choose the right models depending on its strengths and your needs. Some models excel in code generation, while others are more effective for high-level reasoning and planning. Sometimes you need greater context limits, and other times you might need a cheaper model, especially for simple routine tasks.

Closely related to model choice are data privacy settings. Cursor provides controls that let you manage this under Cursor Settings > General > Privacy Mode. You can explicitly choose to restrict future model training on your code if you are working with sensitive, proprietary code.

A quick check of your keybindings is also essential for maintaining an efficient workflow. While Cursor inherits all standard VS Code shortcuts, it introduces new AI-specific commands that you will want to commit to muscle memory. You can access the Keyboard Shortcuts editor to view, modify, or override any key combination.

Cursor rules

Cursor uses a structured system of rules to maintain code quality and act as a reliable partner. These guidelines fall into three main categories:

• User rules for your personal preferences,

• Project rules for repository-specific standards,

• Team rules for organization-wide governance.

By defining these, we prevent the LLMs from generating code that violates established styles, security practices, or architectural patterns.

Project rules are stored locally within your repository, specifically in the .cursor/rules directory. These are created as .mdc (Markdown with context) files. A file contains the rule description and essential metadata, such as glob patterns (globs), which tells Cursor exactly which files or contexts the rule applies to.

You can easily create these rules directly under Cursor Settings > Rules, Memories, Commands. Here’s what a project rules file might look like:

For organizations, team rules are managed centrally through the Cursor dashboard and distributed via Mobile Device Management. Cursor follows a strict order: team rules take the highest priority, followed by project rules, and finally user rules. This hierarchy guarantees that critical organizational standards always supersede specific project configurations or personal preferences.

Additionally, you can utilize agent instructions. These are configured in your project root as AGENTS.md. These guide the high-level behavior of the agent and require no metadata. You can also organize instructions in subdirectories, allowing you to apply specific constraints only to that folder's scope:

application/
  AGENTS.md              # project-wide rules
  services/
    AGENTS.md            # Service-layer instructions
    auth/
      AGENTS.md          # Authentication module constraints
    payments/
      AGENTS.md          # Payments module best practices
  infrastructure/
    AGENTS.md            # Infrastructure-layer requirements

Conclusion

This topic introduced you to the foundations of Cursor. As an AI-native IDE, it is designed to integrate LLMs directly into your coding workflow. We walked through the installation process and explored the interface. We distinguished between the standard Editor view for manual coding and the Agent view for AI-assisted problem-solving. We also covered essential configurations, such as managing models, ensuring data privacy, and customizing keybindings to fit your habits.

Finally, we examined the Cursor rules system, which ensures that Cursor acts as a reliable partner. You learned how to establish a hierarchy of rules using .mdc files to enforce your specific coding standards. With these core elements configured, you are now prepared to use Cursor to write cleaner code and debug more efficiently.