Typically, a Starsim run will result in an output which is a file, or a series of files with names like gstar.1.fz, gstar.2.fz etc. Regardless of whether we run locally or on the Grid, there is a small chance that the file(s) will be truncated. To guard against the possibility of feeding up incorrect data to the reconstruction stage, and/or performing a split or merger of a few file, a KUMAC script has been developed. It will, among other things, discard incomplete events, and produce serially numbered files with names like rcf1319_01_100evts.fzd, which contains the name of the dataset, the serial number of the file (distinct from the numbering of the input files), and the number of events contained therein, all of which is helpful in setting up or debugging the production. It has recently been simplified (although still not easily readable), and wrapped into a utility shell script, which does preparation work as well as cleanup. The resulting script, named "filter.tcsh", takes a single argument which is assumed to be the name of the dataset (and which is then used in naming the output files).