We are trying to use the API to get slow queries from a given cluster. Today, each line represents a single command, but it would be very usefull if we could group it by command shape. This way we could for example group it and filter by average running time, so we could simply get the most expensive queries despite the query data.