Security panels really are a messed up grey area…
From a purely technical standpoint, they are buggy, and full of security holes, when computer interfaces are attached to them (RS-232 is less of a security risk than the Ethernet interfaces), as an example, los93sol and I did a driver for the Honeywell/Ademco Vista 15/20, attached to a Vista ICM module. The Vista ICM module exposed an HTTP interface, and a CCP port which runs over UDP. The CCP port actively broadcast all sorts of useful information about the topology of the sensors it was connected to, and their status, using the subnet broadcast address, anyone who wanted the information simply needed to bind to 192.168.80.255, and wait for the packets to roll in.
But this device is covered by most insurance companies.
There are a few other security panel drivers in LinuxMCE, but I did not write them, nor do I know how to configure them.
Basically, a security panel may have a concept of zones and partitions in order to separate out sensors to make it easier to program different security modes. When you use LinuxMCE's security panel in orbiter, it sets the security mode for the house as a whole, and the device drivers tend to be developed like this. With this said, the security plugin can be passed zones manually as part of scenarios that you manually create, but that's your responsibility.
The other extreme of this, is to simply attach sensors to GC-100's, or to a Z-Wave, or X-10, or PLCBUS, or KNX network, and to let LinuxMCE manage these sensors directly and react to them. This gives you the most flexibility, but it also means that your insurance company will not insure the security system.
-Thom