Here's a simple approach, which just writes information about the instances in the cluster to a tab-delimited plain text file called ~/.whirr/<cluster>/instances. It should be easy to use unix tools to parse this file and operate on instances in the cluster. Here's an example of the file for a Hadoop cluster:
us-east-1/i-54cbe839 tt,dn ec2-67-202-4-185.compute-1.amazonaws.com 10.240.61.238
us-east-1/i-0ecbe863 jt,nn ec2-72-44-46-97.compute-1.amazonaws.com 10.240.62.31
The columns represent ID, roles, public address, private address.