My point is that the presentation-layer is the single source of truth because that is what users see and and act upon. It's the thing that matters and what apps are actually designed around. Thinking otherwise is dismissing the user's own understanding of Nostr data in favor of some engineering purity.

And again, this is not a blockchain. The past is changing all the time. And if you decide to ignore the changes, your app diverges from rest of Nostr.

Correct. But in practice, in all nostr apps I have seen, the information within them is mutable for the user. Which creates as many problems as usual native mutable records. The base data structure being immutable is completely irrelevant in real world apps. I call this the developer fantasy for immutability.

And remember, nostr is not a blockchain. Things can be deleted and recreated, with dates in the past and the future. That is also a way to mutate records (and the mindset of users).

Immutability in Nostr is a pipedream, an illusion. It's fools gold.

I think agentic bug finding is easy and needed. But I agree that the most important thing we do is to evaluate if the feature and UI makes sense. I am discarding a lot of things the Claude does not because it is wrong or not working, but because it's doesn't make sense to the user.

Until agents can run and walk through apps to test their own features..

Then maybe if people add them as a blossom server, amethyst already uploads there for the feed.

They still suck for reviews :(

Does it use a blossom API?

Not even if you know which npub owns it? Trust (and pass the liability to) the user, not each file. You can enable only for paying users for instance...