One could argue that you should never consider any input as “safe to consume”, and if you write defensive app code, would never consider the ENV to ever be secu

One could argue that you should never consider any input as “safe to consume”, and if you write defensive app code, would never consider the ENV to ever be secure or what you expect, and would therefor sanitize that input before consuming. If you’re blindly trusting any input just because you’re single tenant, you’re in for some hard to track down bugs and long nights when those circumstances one day change due to random business pivots.