During some testing, execution of the entire software was interrupted due to unhandled error after encountering an error in a conversion process. I suspect it's due to incorrect stream handling (converters already tinkering with the stream, probably closing it, and then something else, probably in CLI, still trying to operate on it). Haven't looked into it back then but I guess it's something that should be looked at - just in case someone will try to throw in some invalid files into the software.