AI Prompt Writing Guide

Prompt Writing Basics

Writing effective prompts is key to getting the best results from Cursor's AI features. A good prompt should be:

• Clear and specific
• Contextually relevant
• Properly formatted
• Focused on a single task

Prompt Structure

An effective prompt typically includes these components:

1. Task Description
   - What you want to accomplish
   - Specific requirements or constraints

2. Context
   - Relevant code or background information
   - Programming language or framework

3. Expected Output
   - Format of the desired result
   - Any specific requirements

4. Additional Parameters
   - Performance considerations
   - Style preferences
   - Error handling requirements

Providing Context

Help the AI understand your codebase by:

• Including relevant code:
  • Function definitions
  • Related classes or modules
  • Important dependencies
• Specifying technical details:
  • Programming language version
  • Framework constraints
  • Platform requirements

Example Prompts

Here are some effective prompt examples:

"Explain this React useEffect hook:
[code block]
Focus on:
1. Dependencies array usage
2. Cleanup function
3. Potential side effects"

"Create a TypeScript function that:
- Validates an email address
- Returns a strongly typed result
- Includes error messages for invalid cases
- Uses regular expressions
- Has JSDoc documentation"

"Refactor this function for better performance:
[code block]
Requirements:
- Reduce time complexity
- Maintain readability
- Add error handling
- Keep existing functionality"

Common Tasks

Effective prompts for common development tasks:

• Debugging:
  • Include error messages
  • Describe expected behavior
  • List attempted solutions
• Testing:
  • Specify test framework
  • Define test cases
  • Include edge cases
• Documentation:
  • Specify documentation style
  • List required sections
  • Include example usage

Best Practices

Follow these guidelines for better results:

• Be specific and clear in your requests
• Break down complex tasks into smaller parts
• Provide relevant context and constraints
• Use appropriate technical terminology
• Specify output format preferences
• Review and iterate on responses

Troubleshooting

If you're not getting the desired results:

• Unclear Responses:
  • Add more specific requirements
  • Break down the request into steps
  • Provide examples of desired output
• Incorrect Solutions:
  • Check if all context is provided
  • Verify technical requirements
  • Clarify constraints and limitations
• Performance Issues:
  • Reduce prompt complexity
  • Focus on specific aspects
  • Use step-by-step approach