All posts at the time of writing, including the OP, seem to assume that passwords only contain digits; at least, that's all that's being tested. I see nothing at Asterisk Voicemail (or pages that links to) to indicate this restriction. Accordingly, 'a12b543c', for instance, would be valid (as a standard password) but invalid (under your consecutive digits special rules).

In the test script below, sequential digits are pulled from the whole password; canonicalised into ascending order (e.g. 'a12b543c' would produce the two sequences, '12' and '345', for checking); tested for consecutiveness; and rejected immediately any check fails.

#!/usr/bin/env perl -l

use strict;
use warnings;

my @passwords = qw{
    1234 1243 4321 298761 4562 4568 4578 123 12 1
    01234 01243 04321 0298761 04562 04568 04578 0123 012 01
    a1234 1a234 12a34 123a4 1234a
    a1b2c3 a12b34c56 a1b2c3d a12b34c56d
    a123b45c6 a12b345c6 a123b45c6d a12b345c6d
    1a2 1ab2 12ab34 12abc34def 12abc34def567
};
push @passwords, map { scalar reverse } @passwords;
my $too_many = 3;
check($_, $too_many) for @passwords;

sub check {
    my ($pw, $too_many) = @_;

    if ($too_many > length $pw) {
        pw_ok($pw);
        return;
    }

    for my $pw_digit_str (split /\D+/, $pw) {
        my $pw_digit_str_len = length $pw_digit_str;
        next if $too_many > $pw_digit_str_len;

        OFFSET:
        for my $offset (0 .. $pw_digit_str_len - $too_many) {
            my $digits = substr $pw_digit_str, $offset, $too_many;
            my $rev_digits = scalar reverse $digits;
            my @ints = split //, $digits < $rev_digits ? $digits : $re
+v_digits;
            my $test_int = $ints[0];
            for (@ints) {
                if ($test_int != $_) {
                    next OFFSET;
                }
                ++$test_int;
            }
            pw_nok($pw);
            return;
        }
    }
    pw_ok($pw);
}

sub pw_ok  { print "Accept: $_[0]" }
sub pw_nok { print "Reject: $_[0]" }
[download]

This generates 76 tests (I did note a couple of duplicates). The output is in the spoiler, below.

I've covered many edge cases. Don't assume I've caught them all.