First draft. Seems the right idea if not a final. Probably needs some kind of a "binary truncation" to avoid doing a crap ton of work if given a couple megabytes of character data but only looking for the first 255 bytes.

Update, redacted the update, it was wrong and only up for 30 seconds. :P

use 5.16.0;
use strict;
use utf8;
use open ":std", ":encoding(utf8)";
use Encode;

# my $str = "艾捷克, 萨塔尔, 胡西它尔.";
my $str = "艾捷"; # Shortened string  for terse output example.

my $max_bytes = shift || length encode(utf8 => $str);

while ( $max_bytes )
{
    my $length;
    $str =~ s/.\z// while ( $max_bytes < ( $length = length encode(utf8 => $str) ) );
    $length ||= 0;
    say <<"";
        Max -> $max_bytes
     Actual -> $length
     String -> $str

     $max_bytes--;
}
__END__
        Max -> 6
     Actual -> 6
     String -> 艾捷

        Max -> 5
     Actual -> 3
     String -> 艾

        Max -> 4
     Actual -> 3
     String -> 艾

        Max -> 3
     Actual -> 3
     String -> 艾

        Max -> 2
     Actual -> 0
     String -> 

        Max -> 1
     Actual -> 0
     String ->