Randomly overengineering things with AI

I just thought about something and decided to write a short blog about it.

AI has made it incredibly easy for me to go to my old projects and overengineer the hell out of them.

For instance, I have a repository where I store all my containers for development and rpm-ostree usage: https://github.com/ryanabx/ryanabx-containers

Now that I have a local LLM that is essentially free (minus electricity), I can just get it to do random stuff.

• I had it build me a ryanabx-kinoite image that does the same thing as my ryanabx-dev image but in ostree form.
• I had it resolve all warnings in my GitHub Actions CI, most of them just being deprecated actions I was calling into
• I had it revamp the README and make it pretty (and of course it got at least one thing wrong I had to correct: the command for how to rebase with rpm-ostree was missing, instead it recommended to use an ostree admin command). I even told it to add those stupid shields.io badges because why the heck not?
• I had it set up container signing, which I don’t even know why I’d need but why not?
• And just now, I think I’ll try to get it to create an SBOM, because again, why not?

I’m finding that for all the big things AI has helped me with at work and with home development, sometimes the little things are interesting because of how AI can make them ridiculously overengineered.

Anyways, that’s all my thoughts for now. Happy Friday!