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.

Corion,
I get an empty string when I print the file content.

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

Here is the result returned by the Data::Dumper

$VAR1 = bless( {
                 '_headers' => bless( {
                                        '::std_case' => {
                                                          'x-frame-opt
+ions' => 'X-Frame-Options',
                                                          'expect-ct' 
+=> 'Expect-CT',
                                                          'content-sec
+urity-policy' => 'Content-Security-Policy',
                                                          'x-xss-prote
+ction' => 'X-XSS-Protection',
                                                          'x-content-t
+ype-options' => 'X-Content-Type-Options',
                                                          'strict-tran
+sport-security' => 'Strict-Transport-Security',
                                                          'referrer-po
+licy' => 'Referrer-Policy'
                                                        },
                                        'content-security-policy' => '
+default-src \'self\' data: https: \'unsafe-eval\' \'unsafe-inline\'',
                                        'date' => 'Sun, 11 Jul 2021 19
+:46:11 GMT',
                                        'strict-transport-security' =>
+ 'max-age=31536000; includeSubDomains',
                                        'etag' => '"06c81313776d71:0"'
+,
                                        'expect-ct' => 'enforce, max-a
+ge=30, report-uri="https://{$subdomain}.report-uri.com/r/d/ct/enforce
+"',
                                        'x-frame-options' => 'SAMEORIG
+IN',
                                        'server' => '',
                                        'x-content-type-options' => 'n
+osniff',
                                        'x-xss-protection' => '1;mode=
+block',
                                        'accept-ranges' => 'bytes',
                                        'referrer-policy' => 'no-refer
+rer'
                                      }, 'HTTP::Headers' ),
                 '_request' => undef,
                 '_content' => '',
                 '_rc' => 304,
                 '_msg' => 'Not Modified'
               }, 'HTTP::Response' );
[download]

Corion,
It looks like the issue is related to the file type. If the .csv file is replaced with .html file, the ->get() returns the content of the file.

Corion says: Did you set the download directory (download_directory option) in the constructor?
-----------

The download_directory finally worked after finding out that chrome does not support paths in this format c:/path...
The download path has to use the backslash instead ( c:\path... ).

my $downloads =  = "C:\\path\\";
$mech->set_download_directory( $downloads);
$mech->get($foo);
[download]

My final question is how to stop chrome browser from loading some documents so that the download can be executed with WMC. If a document (such as a jpg) is loaded in the browser, the file does not download.
Thank you for your patience.

If a file does not download, have you tried inspecting the HTTP::Response object you receive from the ->get() call?

my $response = $mech->get($url);
open my $output, '>:raw', '/tmp/output.jpg';
print { $output } $response->decoded_content;
[download]

Edit: You might need to touch the ->content of the browser first so everything has time to initialize first:

my $response = $mech->get($url);
my $c = $mech->content; # Dummy request to initialize everything
open my $output, '>:raw', '/tmp/output.jpg';
print { $output } $response->decoded_content;
[download]