Vega refers to indices instead of Data Views. For this reason, it does not automatically inject the runtime_mappings for any Runtime fields created in a Data View.

It may result in some visualization issues when the Vega visualization is added to a dashboard and the user adds a filter that depends on a Runtime field added to the Data View (runtime fields added to the mappings of the index don't have this issue). The issue is caused because the filter is spread to the Vega visualization, but it is requesting ES to filter the content by a non-existing field, resulting in Vega getting empty results.

While this is technically correct, it may cause some frustration as users may want to their Vega visualizations to follow up with the filters and update appropriately. As a workaround, they can copy the runtime_mappings declaration to their Vega configuration. However, it'd be great to have a custom %variable% to tell Kibana to automatically inject the runtime_mappings to their query.