What would really be great for a filter would be to be able to return all tagged lines from all searched files with a few lines of context.
With a few other features (or maybe as a parallel product) this could become a reasonably priced alternative to qualitative data analysis (QDA) software costing $500-1000. 90% of what people do in these packages is paragraph-level coding (of interview transcripts etc.) & a program which could filter multiple text files by tag/code would cover that.