Or even have a prompt work on multiple files --- why isn't there an LLM front-end which will accept as input a folder of files, and then run a working/tested prompt on each file in that folder, then return the collected output as a single process/result?