Using a URL based data source requires that the URL return the results JSON format
The query itself inside Redash will simply contain the URL to be executed (i.e. http://myserver/path/myquery) and result with all the data it has - the SQL equivalent would be
SELECT * FROM url.
Here's a list of valid column types returned in results:
To manipulate the data you get from your URL data source you can save the query that resulted in all the data and query that data set. Read more about querying query results here.