How It Works
With a internal
/api/__kql server route proxying your query request.
The internal server route then fetches the actual data for a given query from the Kirby instance and passes the response back to the template/client. Thus, no KQL requests to the Kirby instance are initiated on the client-side. This proxy behaviour has the benefit of omitting CORS issues, since data is sent from server to server.
During server-side rendering, calls to the Nuxt server route will directly call the relevant function (emulating the request), saving an additional API call. Thus, only the fetch request from Nuxt to Kirby would be performed.
Query responses are cached (with a key calculated by hashing the query) and hydrated to the client. Subsequent calls will return cached responses, saving duplicated requests.