The API supports a notion of custom data “combinators”, with the key feature being data transformation.

It allows for: * combining data from multiple feeds into a single response stream * remapping data JSON from such different streams into structures chosen by the developer to facilitate consistent structures across unrelated sources * ordering of data according to underlying JSON structure fields * filtering of data according to underlying JSON values (including text-based search)

Currently it is achieved by dynamically registering a datapoint with a data mapping specification and GETing data from the registered endpoint