Sequential IDs make submitted code excessively non-private.

[MyrddinE]

I'm aware that this is a tool that allows anonymous users to compile and run code, but I think that a more appropriate method would be to generate random IDs rather than sequential to prevent visitors from browsing the code submitted by others.

If one-in-a-billion IDs was valid, it would make submissions functionally private even if they're still technically public.

[MyrddinE]

Looks like the @id is generated by the database sequentially, then just converted to base36.

Perhaps using PostGres' gen_random_uuid() function? It would guarantee uniqueness, but it would make the URL a lot longer. Some increase is inevitable if switching from sequential to random. It's almost certainly the approach that would require the least work.