I need to filter outgoing mail for spam and have come up with an approach. I'm wondering if anyone else here has implemented a solution to this problem, and am posting to let you all shoot holes or suggest improvments. :).

1. Create separate smtp server that only webmail can connect to.

2. On this courier instance, cat > esmtproutes :[ip-of-current-outgoing-mail-server] ^D

3. Wrap Vipul's Razor in Courier::Filter module.

4. Reject mail if Vipul says so.

This approach assumes courierfilter will run before esmtproutes gets evaluated and used?

Is this assumption correct?

As a related question, do folks here add to Vipul's checksum database in any kind of bulk way? If so, how? I was thinking I could pipe any mail that got a hit on cbl.abuseat.org to a daemon that reports it as spam to Vipul. Not great, but I guess I would trust cbl more than user training of dspam. I could just do my own spam, but that seems wimpy given the vast volume of spam flowing to our mail servers.