Querying
Running SQL from the console
SQL Workbench runs queries against your databases from the browser, with a catalog of your schemas and tables beside the editor.
Queries run on your login, not the database's
This is the part worth understanding before you use it.
A query you run in the console goes through the HiveOps API, authenticated as you, the signed in developer. It does not connect to the database with an app user's credentials, and it does not need the database to be reachable from the internet.
Two consequences follow.
Turning internet access off does not stop you querying from the console. The console reaches the database from inside the cluster.
An app user's capabilities do not limit what you can run in the console. Those capabilities apply to that login, connecting over the network. They are not a sandbox around you.
What it is for
Reading data, checking a migration, and the ad hoc questions that come up while building something.
It is not the path an application should take. Applications connect directly with their own app user, which is what Connecting covers.
Analytical queries
A transactional database is built for queries that touch a few rows and return quickly. A query that scans the whole table will work, and will slow down everything else while it does.
If that is the shape of your work, the Zero-ETL mirror exists so heavy queries never touch the database serving your users.