Vibe improvements

This image was created by Chat GPT with the prompt 'Create a cool looking, minimal hero image for a blog post about vibe coding'

This image was created by Chat GPT with the prompt 'Create a cool looking, minimal hero image for a blog post about vibe coding'

Using AI to “vibe code” some improvements to this place after a good number of years of neglect.

Here are some of the projects that Github Copilot and Anthropic Claude have helped me accomplish, with no more than an hour devoted to each:

  • Added a light/dark theme switcher to the entire site, that respects a user’s system setting and allows them to override it
  • Added micro blog posts into the main blog posting feed
  • Added top 5 directors list to the Movies rating page
  • Added top 5 years and top 5 systems to the Video Games page

(I also restyled the ratings items, but I didn’t need - or want - AI help for that.)

So… AI.

It’s been fun using these crazy tools. Seeing what they can do with so little given to them in terms of prompts, watching them learn from the files I upload to them - it’s all really fascinating.

A case study

Also, a file that I usually work on annually for my playoffsbracket site is one that usually takes me hours to complete by manually referencing and copying and pasting NBA scores into a custom format I established several years ago. I also needed to copy and paste the URLs for each game I reference in the playoffs from nba.com to that file. This is a slog for me that I was actually planning on putting off this year.

Instead, I decided to show Anthropic Claude my file from last year, and my empty file from this year, and gave it this prompt:

Using 2024.json as an example, add all of the games from the 2025 NBA Playoffs to 2025.json, with links to nba.com and scores.

…and it created the entire file in 5 seconds. It actually scoured the internet for the NBA Playoffs results and the appropriate nba.com links to the games, looked at and learned from my custom format (which likely doesn’t make any sense since I don’t know what I’m doing writing JSON), and wrote the complete file. I saved it and ran it, and it ran perfectly. I didn’t need to make one change. It saved me hours.

Well, ok then

So if it can do that, what else can it do? Apparently alot, as the above list demonstrates. I’m not a programmer. I’m not a frontend web developer. I’m a practiced web designer and design systems specialist, which has nothing to do with writing code in Go or figuring out how to traverse complex JSON data files and return data from them. I just want to make that data look cool on the way out.

Happiness, for now

So, I’m very happy for now to allow AI to take over the heavy lifting of the “complex” programming aspects of my private projects while I move on to the parts I like to do, like styling with CSS, figuring out what else I want to add, and what features I can make better or more accessible. Though I do want to be careful about having AI do a bit too much, and getting lazy with it, to the point where I’m not really internalizing how something works in CSS or how to apply new features. That’s why I’m writing my CSS manually still.


Anyway, this has been pretty fun for the past few days that I’ve been able to implement it. I have some ideas to run through these tools soon that should push them a bit further, so we’ll see how far we can go.