Code Refactoring Prompt

This prompt helps developers plan and execute targeted refactoring to improve code quality, maintainability, and readability without altering core behavior.

Purpose

The refactoring prompt assists with:

  • Planning focused code improvements in specific areas of your codebase
  • Identifying issues like complexity, duplication, and code smells
  • Creating a safe, incremental approach to refactoring
  • Validating that behavior is preserved after changes
  • Integrating with static analysis tools (optional)

When to Use

This prompt is especially useful when:

  • You need to extend or modify legacy code and want to improve it first
  • Code reviewers have suggested improvements after a PR
  • Quality gates (e.g., Lizard or SonarQube) highlight issues in a module
  • You’re improving AI-generated code before adding features
  • You’re paying down technical debt in a controlled, safe manner

How to Use

  1. Run the prompt: Use PROMPT.md as your prompt to Copilot or another AI assistant.

  2. Answer the initial questions:
    • Which part of code needs refactoring (files, directories, components)
    • Your main concerns with the current code
  3. Let the AI analyze your code: The AI will:
    • Examine your codebase structure
    • Identify potential issues
    • Ask clarifying questions about your priorities
  4. Review the refactoring plan: The AI will propose:
    • A step-by-step refactoring strategy
    • Specific code changes to implement
    • Testing approaches to verify behavior
  5. Execute the plan: Follow the suggested steps, making incremental changes and validating as you go.

MCP Integration

This prompt supports Model Context Protocol (MCP) integrations to enhance analysis with:

  • Static analysis tools like SonarQube
  • Complexity metrics from tools like Lizard
  • Linting results from ESLint/TSLint
  • Test coverage data

If these integrations are available in your environment, the AI will automatically leverage them.

Example Usage 

I need help refactoring the authentication flow in my React application. It's currently spread across multiple files in src/common/auth/ and has become overly complex with duplicate logic. I want to improve its structure and testability without changing the core behavior.

The AI will guide you through analyzing and improving the code in a conversational manner.