miniblog.

Agentic programming workflows rather remind me of genetic programming. The agent has a validation step that looks like a fitness function, and both run iterative trials.

I've been really enjoying paru as a pacman substitute on Arch Linux: https://github.com/Morganamilo/paru It allows you to update both normal and AUR packages in one go, which is super convenient. It also shows you PKGBUILD files, so there's still a human audit step for AUR.

Thinking more about the "auto eval definitions" approach, I think it breaks down when debugging. If I want to step through the existing definition, it'd be really awkward to re-evaluate the definition at my cursor.