[SahAssar]

> “Postgres environment variable”

I think most people think of environment variables as being for a whole process. For RLS this can be any GUC variable so it can be per-session, per transaction, etc.

Usually you would set it per transaction (and start a transaction for each request) and I think the important part you are missing to say is they are applied to any joins, CTE's, functions and views (as of the latest-1 version with the right flag).

So you write your schema, write your access (RLS) rules and can then write your queries as if you had access to the whole DB but the only parts you will see is what that user can access.