Fault-tolerance is one of the main problems that must be resolved to improve the adoption of the agents' computing paradigm. In this paper, we analyse the execution model of agent platforms and the significance of the faults affecting their constituent components on the reliable execution of agent-based applications, in order to develop a pragmatic framework for agent systems fault-tolerance.The developed framework deploys a communication-pairs independent checkpointing strategy to offer a low-cost, application-transparent model for reliable agent-based computing that covers all possible faults that might invalidate reliable agent execution, migration and communication and maintains the exactlyone execution property.