Variable mask in the middle of the password

[pkern]

All your findings are correct. There is no "native" way to do such a mask attack.

But even if there's no native way, how would you approach this? If I try prepend/append rules (i.e. a reversed set of ^s and a set of $s) for the whole thing I end up with something that cannot be used by the OpenCL kernel ("Cannot convert rule for use on OpenCL device"). Apparently because I need to specify too many of them.

AIUI I can't use the maskprocessor either if I want to use a dictionary first because that's just brute-force. Even though that at least would easily support prefix/suffix use. Plus I'd need to have some kind of mutation rule on the input anyway. Is the solution really to try to get something to emit lines fast enough to feed into hashcat as a pipe?