We’re capturing explicit individual files, but with significant paths; so we could capture the file and then nest it in a bunch of directories, but that nesting would be a (very easy to write, but not hugely efficient) new API