Did you know AI can ask you questions before writing code, if it’s unclear on anything or have better suggestions? So it’s not purely prompt in -> execute task. It’s the new way of AI prompting.
At VibeCodeCleanup.org, we’ve designed a structured framework to help developers and AI work together seamlessly. The Collaborative Code Generation Framework (V1.0) is a guide for using AI to write code collaboratively, while keeping humans firmly in control.
A lot of people are not aware of this, but for bigger projects and scalable solutions you should be giving AI a whole file to describe context and among other things.
People have been trying to reverse engineer AI through prompting and this prompt file should give you the basics of what you need to generate great results whilst coding.
Just make sure you tell it to read this file before your actual prompts.
Here is the Ultimate Prompt Starter Pack file in TEXT format – feel free to share it and come back and visit (even better share this website!), we’ll be keeping it updated:
Last Updated 15/09/2025
Prompt Starter Pack Breakdown:
1. Setting the Scene
Before any coding starts, we need AI to understand the project context. This ensures it doesn’t hallucinate. We need to give it an overall View of the project and specifically the important files in the project for the tasks we want it to achieve:
- Project Description:
- A clear, concise overview of the project. Here you will define the purpose and state of the project.
- Memory Injection:
- Referencing dependencies like libraries, frameworks, files to main context and consistency across all prompts generation.
2. The Mission
What problem needs solving? Ensuring the AI doesn’t take over and go bananza, we need to clearly state:
- Objective & Description:
- High level goal
- Current State
- If there’s anything pre-existing you’re in the middle of
- Desired Outcome
- What is the end result here – what do you want to achieve in terms of measurable outcome
3. Iterative Checking
Make AI ask you questions – it’s possible you may have had an oversight. It’s also a perfect phase for re-defining your prompt file:
- Directive: Ask it to ask you stuff. You can modify the pre-built directive or leave it as is in the starter pack.
- Condition: If AI wants to continue, what condition must be met
4. Divide and Conquer
Here’s where you define steps for AI to take, yes AI is intelligent but it will wander off and do it’s own thing a lot of the times. Ever notice it say “Hey… while I’m at it I also see X problem, just let me fix it” and the actual problem was something specific you implemented?
So the solution is to provide AI a chain of thought like so:
- Create the main User class
- Add properties and methods
- Implement error handling
5.Technicalities
Here you define how the code should be written. Including styles, standards, naming conventions, security, examples, etc… It may seem like overkill, but it’s important to understand we’re trying to build something scalable here. Not simply built for one purpose. This approach will mean any dev coming in will be able to follow the standards.
If you’re feeling efficient and have documentation on company standards, get AI to generate you the details of this section.
6. Review
Telling AI you’re human and need a sanity check is a good way to get it to produce more clear and passable code.
7. Iteration
In a perfect world, this wouldn’t be necessary, but since AI can make mistakes, we need to be able to correct those. We should also be telling how we want AI to handle it’s mistakes and allow it to learn from it.
Conclusion
By using this framework we can make coding with AI more iterative and detailed. The framework helps:
- Reduce mistakes
- Maintain code quality
- Accelerate development
- Scale
If you think there’s anything we can add, or want to provide feedback, heck maybe you created an awesome prompt from our template! Please get in touch below. We want this to be an open community where ideas are shared and discovering best practices is what we do!
0 Comments