As we use the SaneSecurity definitions with Clam it would be useful if we could score based upon the signature name being hit. i.e. For Clam definitions it would score sufficient to block in all instances but for the SaneSecurity definitions we could tag a lower score which would mean we could still put the message in the Junk folder rather than block as is the case at present.

If this is not an option, perhaps at least have the ability to disable AV scanning on a per-domain basis.