[sandspit]

There are some people who have done things like this. I think the problem it runs into its just a very specific way to do the thing and any engineer with any experience is going to have their own set of preferences. Its going to end up feeling constraining.

Just looking at this list, personally i would instead go with shadcn, trpc, node test for node, drizzle, pulumi, next auth or clerk, etc.

The time saved with a plug and play stack imo will not be significant compared to the long term time spent building on and maintaining the stack, so a developer would be rational to pick and choose based their specific needs.

Also I think typically its more natural to build incrementally than all at once. Like are they certain they are going to need graphql? Maybe they could worry about analytics in a few months?

I think its super easy to become over-invested in your own preferences, and lose site of how complex and varied real world requirements are.