Yes. So what is the value of $foo? Did you print it? Is the loop necessary for the bug to happen, or does the error happen with the hardcoded value for $foo as well?

From the URL you sent to me, it seems that WWW::Mechanize returns absolute URLs, while WWW::Mechanize::Chrome returns whatever URLs are contained in the file. Simple printing of the values can confirm that to you.

Until WWW::Mechanize::Chrome achieves API parity with WWW::Mechanize in that part, consider converting the URL to absolute yourself.

I modified the code to eliminate the loop and I changed the URL to absolute.

my @urls = map { $_->url_abs } @links;
[download]

The error message went away but these calls still don't return anything.

$mech->get($foo, ':content_file'=>$filename);
my $file_content  = $mech->get($foo);
[download]

Why do you expect the following to work for WWW::Mechanize::Chrome? This is not a documented call:

$mech->get($foo, ':content_file'=>$filename);
[download]

I wonder why you say that the following "return anything":

my $file_content  = $mech->get($foo);
[download]

->get() is documented to return a response, so I suggest that you print it, or inspect it using Data::Dumper.