it thinks for longer when you ask for "ultrathink"

agreed, the cognitive overhead comes from language-level mental models with library-level tooling

complexity scales with every new feature and every new developer realistically when a library requires a language mental model

I would love to see that reduced but I wonder how? 🤔