MooseX::App::Cmd and your Command Line Apps

This is the main class:

package App::MyApp;
use Moose;
use namespace::autoclean;
extends 'MooseX::App::Cmd';

1;

From here, you define your commands as such:

package App::MyApp::Command::get;
use Moose;
use namespace::autoclean;
extends 'MooseX::App::Cmd::Command';

has bucket => (
    is            => 'rw',
    isa           => 'Str',
    traits        => ['Getopt'],
    cmd_aliases   => "b",
    documentation => "bucket name to look up",
    required      => 1,
    lazy          => 1,
    default       => sub { croak "bucket name required" },
 );

has object => (
    is            => 'rw',
    isa           => 'Str',
    traits        => ['Getopt'],
    cmd_aliases   => "o",
    documentation => "object to download",
    required      => 1,
    lazy          => 1,
    default       => sub { croak  "object name required" },
); 

sub execute {
    my ( $self, $opt, $args ) = @_;

    my $container =
    ## this would be something an inherited class might 
    ## provide, a database handle, etc.
      $self->object_thingy->container( name => $self->bucket );
    my $object = $container->object( name => $self->object );
    $object->get_filename( $self->object )
      or die "Couldn't retrieve object: $!";

}
1;

Finally, the initialization script:

#!/usr/bin/env perl
use App::MyApp;
App::MyApp->run;

And it's invoked as such:

myapp.pl get filename.txt

This makes your code much much more modular, allowing you to focus on keeping your business logic in one place and inheriting from it anywhere (much like different views in the MVC concept).