Mxguarddog.com I use it for 2 domains very good and almost no spam in my mailbox I have gotten 1 Email that was spam that got through but my local server caught it and moved it to my junk folder. Source: about 2 years ago

I've been using https://mxguarddog.com/ for years and it does a pretty good job at stopping spam. I'd say maybe 1 email breaks through a month but compared to the amount that shows up in the quarantine list it has a very high stop rate. Source: over 2 years ago

> a very odd and theoretical setup Nope, I use https://mxguarddog.com/ for one of my domains. That acts as a spam filter, but also hides the ultimate MX host. - Source: Hacker News / almost 3 years ago

Spamassassin or rspamd or ASSP. All very good. Spamassassin just works, but you need to feed it a good cohort of spam and ham (about 1000 messages each). Most people only send it the spam which will then bias it quite a bit and not have it classify spam as spam unless it's the worst of the bunch. Spamassassin is a filter for your mail server. Rspamd is the same idea. ASSP is highly configurable and a proxy. Source: 11 months ago

I'm currently looking into using a Bayesian spam filter like Bogofilter or Apache Spamassassin to help out. I'd run the contents of my Contact Us form through the filter after the initial validation, and if the contents are below my spam threshold, I would actually send out the contact us request. Source: over 2 years ago

The famous spam filter SpamAssassin. Score: -0.6. A score below -5 is considered spam. -0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid This rule is automatically applied if your email contains a DKIM signature but other positive rules will also be added if your DKIM signature is valid. See immediately below. 0.1 DKIM_VALID Message has at least one valid DKIM or DK... Source: almost 3 years ago

To detect whether incoming emails are malicious, mail servers also use spam scanners, such as Apache's popular SpamAssasin. The internal workings of these spam scanners can be somewhat complicated (involving Naive Bayes Classifiers on trained, large datasets, for the curious), but the primary takeaway is that these classification systems typically assign a numerical point value to an incoming email to determine... - Source: dev.to / about 3 years ago