Vibe Coding Efficiency Tips: A Must-Read Before Writing Code! Use AI to Find the Optimal Solution First.

A friend asked me the other day how to improve efficiency when using Codex.

I told you to connect the GitHub plugin first.

He said he got it connected.

I told you to copy and paste the following prompt and run it before starting each project.

He tried it and came back to tell me, "Holy crap, it saved me so much time."

I've been thinking about it, and it seems quite a few people are unaware of this feature. It's like many people use AI to write code, immediately starting with Vibe Coding, letting the AI ​​generate websites and write apps.

There's really no need to rush.

What you want to do, someone on GitHub has most likely already done.

What you want to do has most likely already been done on GitHub. Many solutions are already mature, and the code quality might even be higher than what you could write yourself. Why rewrite it from scratch when you can directly refer to and modify existing work?

My personal habit is that whenever I want to start a new project, I don't write any code right away. First, I send this suggestion to Codex:

我要做一个XXX。先不要写代码。去GitHub找能直接使用或二次开发的开源项目，确认是否还在维护、部署是否麻烦、哪些功能可以复用。最后告诉我，应该直接用、基于现有项目改，还是自己开发，并给出最简单的MVP方案。等我确认后再动手。

Just this one sentence.

This prompt did three things.

Think about it, this prompt does three things.

First, it allows AI to find existing solutions on GitHub instead of starting to write its own. Many people may not know that Codex, when integrated with the GitHub plugin, can actually search for open-source projects in real time; it's not just a feature.

Secondly, it uses AI to help you evaluate the quality of these projects: whether they are still being maintained, how complicated the deployment is, and which features can be directly used. If you were to manually search through each project on GitHub, it might take an hour or two. Let the AI ​​run it, and you'll get results in minutes.

Third, it allows AI to give you a clear decision suggestion: whether to use it directly, modify it, or write it from scratch, and it also helps you plan the simplest MVP solution.

Why can we save a lot of tokens?

Let me tell you, this operation can really save you a lot of tokens.

Why? Because if you directly let AI start writing code, it won't know that what you want already exists. It will generate a huge amount of code for you from scratch. This code may overlap with the functionality of existing open-source projects, and you'll ultimately have to clean it up yourself.

However, if you let AI help you find the answer first, its suggestions will be based on the existing ecosystem, and when you write the code, the amount of code can be reduced by at least half.

That's what I do myself. Whenever I start a new project, I run this suggestion program first, and then decide on the next step based on the AI's suggestions. It really improves efficiency.

To be honest, I'm not sure if this method will work for everyone; some projects may indeed lack readily available solutions. But think about it: for most of the things you want to do, with hundreds of millions of repositories on GitHub, there's a very high probability that someone has already done it.

Standing on the shoulders of giants

You're not a programmer, so you don't need to write code from scratch. You're not a content creator, so you don't need to create original content every day. You're just someone who wants to do something meaningful, and the smartest approach is to first see what others have already done, and then stand on the shoulders of giants.

Sometimes I feel that the smartest way to use AI in the AI ​​era is not to let AI do everything for you, but to let AI do information retrieval and decision analysis for you, and then you can execute it.

What's the difference between this and using a search engine before? Previously, when you searched for a specific requirement, a bunch of links would appear, and you had to click through them one by one to see which ones worked and which didn't. Now, you tell the AI ​​your requirements, and it filters them for you and gives you the conclusion directly.

Can you believe it?

This prompt should be engraved.DNA:in

I personally think this prompt should be ingrained in our DNA . Running it through before starting a new project really saves a lot of time.

Another advantage of this approach is that you avoid reinventing the wheel. AI helps you find readily available solutions, allowing you to focus on the truly customized aspects. This way, your energy can be devoted to the most valuable areas.

My own experience is that when I first started using AI to write code, I was really excited and wanted the AI ​​to generate everything from scratch. But after using it for a while, I found that doing so was often slower. This is because you have to debug, modify, and maintain the code generated by AI, while existing open-source projects have been verified by many people, so their quality is more guaranteed.

This might sound counterintuitive, right? But think about it: the most important thing about code isn't writing it, but writing it correctly. Existing solutions have already been validated; using them directly will significantly reduce the chances of encountering problems.

So my advice is, next time you're writing code with Codex or any AI, don't rush into it. First, run the prompt once and let the AI ​​help you find the answer.

You might find that what you're trying to do has already been done by someone else.

You just need to take it and use it.

Use the time you save to do something more interesting.

Since you've read this far, if you found it helpful, please like and share it. If you want to receive updates first, you can also follow me!

Thank you for reading my article. See you next time.