I honestly think the given answer is weak. There is no reason a valid query could not, and should not be saved into the index. It is perfectly reasonable for someone to build an application outside of kibana but want to leverage the search creating functionality of kibana. Off the top of my head, any application that needs to do heavy ML processing of NNs across GPUs would be hard pressed to do it with node via the kibana dashboard. Python and TensorFlow are much more suited. It is also reasonable to think the user would want to automate it by some method of retreving the RESULTS of the "saved searches. Whether that is by getting a valid Elasticsearch query or through a REST endpoint that returns results.
Its impossible currently, because we have to "trust" the APIs for how the query is saved in kibana, won't change for new ways of displaying the data in kibana. Seems to me kibana should be saving the raw query into elasticsearch, or at the very least, expose an endpoint that returns it when requested.