In a world where AI-generated code is becoming the norm, managing and tracking your prompts is essential. That's why I built prompt-keeper: a lightweight CLI tool to help developers trace their AI-generated code by storing the prompts alongside their files.

🌐 Check it out:

📦 npm package
🛠️ GitHub repository

🌱 The Origin Story

While developing software, I realized that the prompts used for generating code can significantly impact the code's structure, performance, and maintainability. As AI models evolve rapidly, understanding which prompts and models were used to generate certain code segments becomes crucial.

Moreover, AI code generation is becoming an integral part of modern development workflows. Yet, we often lose the context of why certain code was written a certain way when we rely on AI tools. prompt-keeper solves this by:

Recording prompts linked to code files
Tracking the AI model used
Making prompt management effortless

And yes, I built the first version in just one day! 😎

🚀 What is prompt-keeper?

prompt-keeper is a CLI tool that:

📝 Tracks AI Conversations: Keeps a history of all prompts used in your codebase
🔍 Ensures Code Traceability: Links prompts directly to code files with unique IDs
🤖 Supports Multiple AI Models: GPT-4, Claude, Gemini, and more
🎯 Is Developer-Friendly: Intuitive CLI with interactive and command-line modes

🛠️ Installation

npm install -g prompt-keeper

⚡ Quick Demo

Imagine you're adding a memoized component and using AI to assist you. With prompt-keeper, you can track that interaction like this:

pk i MyComponent.tsx

The tool will prompt you to enter your prompt, select the AI model, and choose where to insert the prompt-id comment. The result?

Code:

import React, { memo } from "react";

// prompt-id: 8ffd78c0-d00c-433d-a8d5-af111cc79020
const MyComponent: React.FC = () => {
  return (
    <div>
      <h1>Test Component</h1>
    </div>
  );
};

// prompt-id: b19e27f8-9b25-4c2f-a011-da3898206f82
export default memo(MyComponent);

Corresponding MyComponent.prompt.json:

{
  "file": "test/MyComponent.tsx",
  "prompts": [
    {
      "id": "8ffd78c0-d00c-433d-a8d5-af111cc79020",
      "prompt": "This is a test prompt for managing AI-generated code.",
      "model": "gpt-4o",
      "createdAt": "2025-02-15T02:00:22.206Z",
      "version": 1
    }
  ]
}

Now, whenever you revisit the code, you know exactly why that component exists and which AI model helped build it.

💡 Why You Should Use It

✅ Historical Context

Ever wondered why certain code was implemented a certain way? prompt-keeper shows you the exact AI prompt behind the logic.

🚀 AI Model Evolution Tracking

AI models evolve, and newer versions often generate cleaner, more efficient code. prompt-keeper helps you track which model was used so you can revisit code and upgrade prompts when needed.

⚙️ Simplified Debugging

Prompts act like documentation. Instead of guessing what a function does, you can trace the initial requirements via the saved prompt.

🚀 Introducing prompt-keeper: Manage Your AI Prompts Like a Pro