I did a quick test and can verify than the add-on is working as intended. However, as expected there are some rough edges around applications that consist of multiple members.

Correct me if I am wrong: The current behaviour is that each filter is applied to every member of a group and the group is only shown when each member satisfies the filter. One can argue that this can make sense for some filters (rio, level, region, ...) but for others it almost never makes any sense (class filters).

I think what would solve this problem without having to change the actual keywords is lifting the filter from propositions to predicates by introducing quantifiers.

Example
I look for the following: A demonhunter and a healer. I would formulate the following predicate:

FA((dh or healer) and rio > 1800) and EX(rio > 2200) and members < 3

EX and FA are the existential and the for all quantifier. dh, healer, rio are bound variables and members is unbound (keyword does not exist yet). Brackets can be used to specify the scope of a quantifier.

You may noticed that we did not specify which variables are bound at the quantifier. This can be implicit. Every keyword that describes a member of a group (every keyword as of right now) is bound and must be quantified. Keywords that describe the group as a whole (members) are unbound and not allowed to be used under a quantifier.

It took me a while to wrap my head around your request, but I think I got some primitive first-order logic implemented as requested. Have a look at the README. Looking forward for your testing results.

Sweet I will test tomorrow.

Sorry that I never reported back. The commit with the quantifiers seems to brake all functionality. Meaning that no matter what I enter all applicants are still there. Nothing is filtered. I think we should discuss that further in #2. Maybe we should close here.

Please continue to give feedback. I pushed version 0.4 which should fix the problems.