Yesterday, I posted about @gigi's vibed nostr client, Feather #1598968

I'm not quite ready to install it on my tester phone, but I've usually been pleasantly surprised by Gigi's sense of style and taste. He may be an example of someone who is doing the vibe-coding thing better?

Now, here is a link to a post about the tools he used to make the nostr client. It is also vibed, so don't feel compelled to read it all, but it is interesting to see the different things he used to make the app -- including his voice: apparently he did not touch a keyboard in making the app.

The thing I'm most curious about is how he does review:

Gigi has scheduled jobs that periodically check for bug reports, issues, and feature requests, then work on them and prepare pull requests.

He says that he reviews the app and retains final decision-making power:

He keeps that decision for merges as well as releases.

But I wonder how real that is: if you are writing an app, and your only interaction with the code for the app is mediated through an agent, I think you probably lose more of the decision-making power than you think. Of course, you can keep the git keys and only merge what you want, but that just moves the power a level up -- you can only merge or not merge what the agent produces.

I feel that I'm making a weak, possibly ludditic argument here, but my own experiences with vibe-coding make me think that I really don't have as much "final decision-making power" as I think.