alt Hacker NewsI'm surprised there is no breakdown of "with skills" (framework) and without. In my experience, apart from human readability, the ability of a model to follow strict skill rules is the important. For example I see a lot less waste of tokens and reasoning retry loops of obvious errors when using Python with uv+ruff than without.
I added pyright hooks to any edit commands in the code base and it works very well to keep it in the rails. Only trick is to set the unused import as a warning due to the way it edits files. Generally, type checks really help with getting the ai outside of its window vision when doing partial edits in the files.