beyti

one of my daily fun routines: - wordle - enclose.horse

As far as I understood, it creates an AST from the whole project and stores it in a sqlite db. Not just the active files. Tries to be a bridge between file system operations of llm models and the disk.

tho, I don't think this will help on context-wise persistence. This is basically a AST wrapper around io operations on disk.

I keep finding myself to minimize the words to describe the same thing as well, since we are finding ourselves doing "that" operation more and more over time.

maybe shortening the term to "auto-code" would help tho.

Masstransit and Mediatr are a strong base I build a lot of my projects, hopefully they will find a sustainable way going forward.

isn't making it paid service will greatly limit its training data in the long run? am I thinking too narrow here?

That's a really clever strategy and so obvious that now I'm feeling guilty of not thinking it out myself earlier. Thanks a lot.

Same here, that would be a really useful thing for my organization.

related: Dunning–Kruger effect