I always tell my LLMs: "Try not to sound like an LLM. And no emojis!"

Works alright.

The whole repo is fucking boilerplate.

It's certainly a sign of something. Not positive at best neutral. As you say it's at best an indication that the author doesn't like writing.

Could it be an indication that the author didn't write the actual code? Is it a sign that the author doesn't really care that much about their project and furthermore could that be a sign that the project is also be be valued by us as much as the author? Maybe the code quality and documentation is less important than the utility. After all many of us don't like writing tests for code!

Perhaps but perhaps we just need to get used to these signs too and get over it.

Then get some freaking taste in READMEs. We can guide the LLMs to better results.

It used to be "don't try to outsmart the compiler", I'm waiting for the time people start saying the same thing about LLMs.

This argument is getting old. Just because you don’t like something, it doesn’t mean everyone agrees and will take the same shortcuts you do. Fortunately not everyone in the world has the same disregard for their own work, and many of us understand the signal it sends when you’re unwilling to even write your own instructions.

If you want your READMEs sloppily written by LLMs, that’s your prerogative. Just like it’s the prerogative of everyone visiting your repo and bumping into a slop README to decide if they want to even give your tool a second look before abandoning it.

Slop READMEs suggest slop code. Soon everyone who’ll even look at your code are other sloppers and (if it ever gets popular) malicious actors who’ll exploit it in an afternoon because no users understand anything the code does.