How can I use SpamAssassin to detect spam?

SpamAssassin scanning should be done in filter_end, after the full message is assembled. spam_assassin_check returns hits, required threshold, triggered rule names, and a report. Use md_spamc_init / md_spamc_check instead when using SpamAssassin 4.0.1+ in client mode.

use Mail::MIMEDefang::Antispam;

sub filter_end {
    my($entity) = @_;

    my ($hits, $required, $names, $report) = spam_assassin_check();

    md_syslog('Info', "SpamAssassin: score=$hits required=$required rules=$names");

    if ($hits >= $required) {
        # Tag the message as spam and add score header
        action_change_header('X-Spam-Flag',   'YES');
        action_change_header('X-Spam-Score',  "$hits / $required");
        action_change_header('X-Spam-Report', $names);
        md_graphdefang_log('spam', $hits, $RelayAddr);

        if ($hits >= 10) {
            # Reject very high-scoring messages outright
            return action_bounce("Message rejected as spam (score $hits)");
        }
    } else {
        action_change_header('X-Spam-Flag',  'NO');
        action_change_header('X-Spam-Score', "$hits / $required");
    }
}