You haven't specified which version of CentOS you're using. The example below assumes Centos 6.7, using the netinstall ISO image to provide network connectivity (via DHCP) during the install. By installing over the network, it avoids the need to swap discs. It's not the only way around your problem, but it's easier than trying to rebuild the CentOS ISO images with local repos.
Here's some sample lines from our Kickstart configuration, which we use to hit a mix of local mirrors of the CentOS Base & Updates repositories, and our own custom repos:
url --url http://reposerver.intranet/repo/centos/6/os/x86_64/
repo --name=updates --baseurl=http://reposerver.intranet/repo/centos/6/updates/x86_64/
repo --name=custom-utils --baseurl=http://reposerver.intranet/repo/custom/
repo --name=vmware-tools --baseurl=http://reposerver.intranet/repo/vmware-tools
The "url" directive (more info) causes Anaconda to pull all base RPMs from the URL specified. The next line includes the updates repository, to catch any updated RPMs published since CentOS 6.7 was released. Finally the last two lines reference our own repos, containing some custom RPMs and the VMWare client RPMs.
In your %packages section, specify the packages you want installed. Anaconda will figure out the dependencies. Precede any you don't want with a dash. Include whole groups by prepending with @, eg @Base.
One thing you might want to do in the %post section of your Kickstart configuration is import the GPG keys for the packages in any custom repos you referenced earlier. GPG keys aren't checked during an Anaconda install, but will be once the system is built and running. This step facilitates updates to a system after installation. Here's how we do it:
rpm --import http://reposerver.internal/repo/keys/RPM-GPG-KEY-CentOS-6
rpm --import http://reposerver.internal/repo/keys/RPM-GPG-KEY-custom
rpm --import http://reposerver.internal/repo/keys/RPM-GPG-KEY-rpmforge