Well, yes.

Just now, any regular user can clone the repository and ask an LLM to tune it to his needs.

I never thought about this before, and it hasn’t been mentioned significantly in the vast amount of AI threads I read here. But it’s a really good point as skeptical as I am (in mid-2026) of AI first codebases

▲

zakirullin 22 days ago | parent | next [-]

BTW, this is not an AI codebase :)

It was mostly crafted by hand.

Let's say, I've saved some "complexity space" for LLM to add features on top.

In other words, the project has dumb-simple code right now, and it is ready to hold some amount of "tech-debt" from an LLM.

	▲	finghin 20 days ago \| parent [-]
		Interesting and probably prescient approach, good luck

▲

adam_beck 22 days ago | parent | prev [-]

Tanner Lisnley did this with React. He didn't share it but talked about his experience. Basically, he re-wrote React with only the parts he was interested in. Sort of like his own version of Preact. You might be interested in his blog post about the subject: https://tannerlinsley.com/posts/projecting-react