[rt-users] Spam management?

[Les Mikesell]

We get about a thousand spam messages a month to addresses like
support at our_domain which, of course, go into RT queues.  Our email
system does site-wide scanning with MimeDefang/SpamAssassin but
because the identification isn't perfect we just tag messages
with a special header instead of discarding at that level.  Our
webmaster alias is handled by the mailman list manager which has
a nifty feature where it can hold messages for approval based on
certain criteria including header contents so only one person has
to wade through looking for false positives and the rejected items
don't clutter the archives.  Has anyone worked out a similar scheme
with RT to require approval on tagged email before allowing it to
go into a queue?  Because of performance problems with large RT
databases I'd prefer to avoid adding the contents and especially
auto-creating the sender as a user as you would have to do if you
just divert into a spam queue (unless there is some way to really
delete the rejected contents).  I suppose it would work to send email
through a mailman list with the RT address as the only member but
that seems pretty cumbersome just to get an approval mechanism.

---
  Les Mikesell
    les at futuresource.com