If the original input data (before your Jolt transform) looks something like this... { "headers": [ "header1", "header2", "header3" ], "rows": [ [ "row1", "row2", "row3" ], [ "row4", "row5", "row6" ] ] }   ...then it may be easier using SplitJSON and EvaluateJSONPath processors in this scenario.   The GenerateFlowFile processor has some sample data that matches the above format using books as the example. The result will be a separate JSON file for each array within "rows". You can configure the MergeContent to bundle the JSON records as needed.     EvaluateJSONPath   SplitJSON   EvaluateJSONPath   ReplaceText     ... View more