Brief Overview
In this blog post, I'll share my thoughts on the growing popularity of Spec Kit.
Why do I like AI tools?
Did you get chills when you used ChatGPT for the first time? Did you go further and start experimenting with GitHub Copilot, Cursor, Gemini CLI, or any other AI tool with "Agent" mode?
If your answer is "yes", you probably know why I like AI tools. The ease of coding that AI assistants provide is just incredible. As I mentioned in my previous article, I was able to implement a complete comment system in just under 2 hours using GitHub Copilot (full story is here). That would not have been possible 2 years ago.
Why do I dislike/hate AI sometimes?
You want something simple. You have a clear goal in mind. You talk with AI. It builds some code for you. But it’s clearly not what you wanted. You give it another shot. Still not better. Third try—final attempt. That’s it.
Ok. Now it’s time to build a whole feature around that small thing. You write a long prompt. You provide all the context you can think of. You even add some code snippets to help the AI understand your codebase better. You wait 5–10 minutes. It’s ready. Quick manual testing… and you realize you’ll waste a lot of time explaining where the AI made mistakes. You’re frustrated.
What now? Should you refine the current implementation? Or maybe start from scratch?
If you start from scratch, you need to explain all the context again. Of course, you can ask the AI assistant to clarify by asking yes/no questions—but will this be enough?
On the other hand, if you refine the current implementation, there’s a chance you’ll find yourself in a loop where every next prompt fixes something, but breaks something else.
Have you been in such a situation? I have. Many times 😑. I think no further explanation is needed.
What problem does Spec Kit solve?
Spec Kit attempts to break down the feature-creation process into small, manageable pieces. Each piece is a progressively more detailed description of the feature you want to build. You can start with a high-level description and then go deeper into details.
That way, both you and the AI can focus on one piece at a time. You can also easily review and refine each step before moving on to the next one.
Spec Kit provides a set of templates to guide this process. You can also create your own templates to customize it further.
By design, each feature lives in its own branch, making it easy to document the vision, the implementation, and the decisions made along the way. This way, the entire decision-making process is captured.
What is the hidden cost of using Spec Kit?
Time. You need to spend a lot of time reviewing and fixing the documents generated by Spec Kit.
This isn’t necessarily a bad thing. You need to understand what you want to build. You need to understand how to build it. You need to evaluate the trade-offs. You need to consider the risks.
This effort makes Spec Kit a very good tool for big and complex features. But if you’re building something small and simple, it might be overkill 🤔.
Time to try it out!
The abnormal growth in Spec Kit’s popularity is reason enough to give it a try. At the time of writing, Spec Kit has over 2k forks (compared to 1.2k just two weeks ago) and 25k stars on GitHub—an increase of around 9k in only two weeks🤯. If that doesn’t convince you, I don’t know what will.
I had a chance to try Spec Kit on this very portfolio website. I’m by no means an expert in using it, but I see the potential. I’ll write a follow-up article once I have more insights to share.
The best thing you can do is try this approach for yourself. If you’re like me and enjoy YouTube tutorials, here is a basic introduction, and here is an explanation of how to use Spec Kit in existing projects.
I hope you found this article useful. If you have any questions or comments, please leave them below.
And as always - happy coding! 🚀