Fixed up, your code would be

use strict;
use warnings;

my $rot = 6;

chomp(my $message = lc(<>));
$message =~ s/([a-z])/
   chr( ( ord($1) - ord('a') + $rot + 26 ) % 26 + ord('a') )
/ge;
print("$message\n");
[download]

or

use strict;
use warnings;

my $rot = 6;

my $src = join '', 'a'..'z';
my $dst = substr($src, $rot) . substr($src, 0, $rot);

chomp(my $message = lc(<>));
eval "\$message =~ tr/$src/$dst/;";
print("$message\n");
[download]

or for repeated use

use strict;
use warnings;

my $rot = 6;

my $src = join '', 'a'..'z';
my $dst = substr($src, $rot) . substr($src, 0, $rot);

my $tr = eval "sub {
   local \$_ = \@_ ? \$_[0] : \$_;
   tr/$src/$dst/;
   return \$_;
}";

chomp(my $message = lc(<>));
$message = $tr->($message);
print("$message\n");
[download]

Set $rot to -6 to "decrypt".

This method of encrpytion is particularly bad because

• the frequency of the letters remains unchanged making the cypher easy to identify and the cyphertext easy to decrypt by observing the statistical distribution of the letters, and
• the punctuation remains unencrypted, making the cyphertext easy to decrypt by observing the statistical length and organization of the words.