Cursor Tutorial

Learn how to use Cursor.sh for intelligent code completion and refactoring.

Getting Started with Cursor

Cursor is an AI-powered code editor that helps you write better code faster. It integrates with various AI models to provide smart completions, refactoring suggestions, and code explanations as you work.

  • Install Cursor from cursor.sh
  • Set up your OpenAI API key or other model provider keys
  • Start coding with AI assistance and explore the features

Compatible AI Models

Cursor supports a wide range of AI models from various providers. Here's a comprehensive list of compatible models:

Model Provider Features/Notes
Anthropic Claude Models
claude-3-5-sonnet-20241022 Anthropic Latest Claude 3.5 Sonnet model with specific version
claude-3-opus Anthropic Most powerful Claude model, excellent for complex coding tasks
claude-3-5-haiku Anthropic Fastest Claude model, good for simple coding tasks
claude-3-5-sonnet Anthropic Balanced performance and speed for daily coding
claude-3-7-sonnet Anthropic Latest Claude 3.7 model with improved coding capabilities
claude-3-7-sonnet-max Anthropic Claude 3.7 with extended context window
Google Gemini Models
gemini-2-0-flash Google Fast response times with good code understanding
gemini-2-0-flash-exp Google Experimental flash model with additional features
gemini-2-0-pro-exp Google Experimental professional model with enhanced capabilities
gemini-2-5-flash-preview-04-17 Google Preview version of next-gen flash model
gemini-2-5-pro-exp-03-25 Google Experimental pro model with specific version
gemini-2-5-pro-max Google Maximum context window and capabilities
gemini-exp-1206 Google Experimental model with specific version
OpenAI GPT Models
gpt-4 OpenAI Powerful language model with strong coding abilities
gpt-4.1 OpenAI Updated GPT-4 with incremental improvements
gpt-4.5-preview OpenAI Preview of next generation GPT-4.5
gpt-4o OpenAI Latest "Omni" model, optimized for coding
gpt-4o-mini OpenAI Smaller, faster version of GPT-4o
Alternative API Models
o1 Anthropic Alternative API for Claude
o1-mini Anthropic Smaller alternative API for Claude
o1-preview Anthropic Preview of newer alternative API
o3 Anthropic Latest alternative API version
o3-mini Anthropic Smaller version of the latest API
o4-mini Anthropic Next-gen mini model via alternative API
Additional Models
cursor-small Cursor Lightweight model developed by Cursor
deepseek-r1 DeepSeek Specialized coding model from DeepSeek
deepseek-v3 DeepSeek Updated DeepSeek model
deepseek-v3.1 DeepSeek Latest increment of DeepSeek model
grok-3-beta xAI Beta version of Grok AI model
grok-3-mini-beta xAI Lightweight version of Grok AI model

Note: Model availability may change as providers update their offerings. You can add custom model names in Cursor's settings.

Setting Up Your Preferred Model

To configure which AI model Cursor uses:

  1. Click on the settings icon in the bottom left corner
  2. Select "AI" from the settings menu
  3. Choose your preferred model from the dropdown
  4. Enter any required API keys for the selected provider

Pro Tip

You can switch between models based on your tasks. Use faster models like GPT-3.5 or Claude Haiku for simple completions, and more powerful models like GPT-4 or Claude Opus for complex problems.

Key Features

Smart Code Completion

Cursor provides intelligent code suggestions based on your project's context, helping you write code faster and with fewer errors.

Press Ctrl+K to trigger AI completion

AI-Powered Refactoring

Ask Cursor to improve your existing code's structure, readability, and performance without changing functionality.

"Refactor this function to be more efficient"

Real-time Suggestions

Get instant feedback and suggestions while you code, including best practices, potential bugs, and optimization tips.

"Explain potential issues in this code"

System Prompts in Cursor

System prompts are instructions that guide how AI models behave when assisting with your code. In Cursor, you can customize system prompts to:

  • Make AI follow specific coding standards
  • Specify preferred frameworks or libraries
  • Set boundaries for AI responses
  • Include project-specific context

Pro Tip

A well-crafted system prompt can dramatically improve the quality and relevance of AI assistance. Be specific about your coding style, error handling preferences, and documentation requirements.

To set a custom system prompt in Cursor:

  1. Click on the settings icon in the bottom left corner
  2. Select "AI" from the settings menu
  3. Scroll to the "System Prompt" section
  4. Edit the default prompt or add your custom instructions

Using .cursorrules Files

A .cursorrules file is a powerful way to configure AI behavior for specific projects or repositories in Cursor. This file allows you to define project-specific AI settings that will be applied automatically when working within that project.

Creating a .cursorrules File

To create a .cursorrules file:

  1. Create a new file named .cursorrules in your project root directory
  2. Format the file using JSON syntax
  3. Define your custom rules and settings

Example .cursorrules File

{
  "systemPrompt": "You are a helpful assistant that specializes in this project. Follow these guidelines:
- Use TypeScript for all new code
- Follow the repository's existing code style
- Use React functional components with hooks
- Document all functions with JSDoc comments",
  "defaultModel": "claude-3-opus",
  "excludeFromContext": [
    "node_modules/**",
    "build/**",
    "*.log"
  ],
  "includeInContext": [
    "src/**/*.ts",
    "src/**/*.tsx"
  ]
}

Key Configuration Options

  • systemPrompt: Custom instructions for the AI when working in this project
  • defaultModel: The preferred AI model to use for this project
  • excludeFromContext: File patterns to exclude from AI context
  • includeInContext: File patterns to prioritize in AI context
  • aiCommand: Custom slash command configuration
  • autoComplete: Settings for auto-completion behavior

Pro Tip

Use .cursorrules files to standardize AI behavior across your team. This ensures all developers receive consistent AI assistance that aligns with project requirements and coding standards.