The skip_rows option will skip valid CSV-rows (and will croak otherwise). Though the hint is good, I think that it doesn't currently do as required by the OP: it skips CSV lines after the header.

When not passing the col_names like you did, it will skip the count of lines as passed like this after reading the header line, so in given example, that would still result in just 3 fields. That obviously needs a new option that would skip lines before the header line in whatever format, so a CSV file like this:

This file holds the report
for sales over August 2013

id,count,description,price
FAIL,,Placeholder,0.00
\N,,,
10D2,4,Drill 9mm,0.80
11E1,1,Festool DRC 18/4.515.15
[download]

could be opened and correctly dealt with with code like:

#!/usr/bin/perl

use 5.20.0;
use warnings;

use DBI;

my $dbh = DBI->connect ("dbi:CSV:", undef, undef, {
    f_dir          => ".",
    csv_auto_diag  => 1,
    });

$dbh->{csv_tables}{stock} = {
    file        => "stock.csv",
    skip_rows   => 2,
    skip_before => 3,  # nonexisting option
    # look ma, no col_names
    };

my $sth = $dbh->prepare ("select * from stock");
$sth->execute;
while (my $row = $sth->fetch) {
    say "@$row";
    }

$dbh->do ("update foo set price = 420 where id = '11E1'");
$dbh->disconnect;
[download]

The new option however has no control over the format of the lines. I'll consider adding a new option for this.