The United Kingdom's Ministry of Defence (MOD) has a large investment in both Computer Generated Forces (CGF) systems and in the supporting data and models that run within them. During the last audit over 20 different CGF were identified in use across the three service domains. A major concern is that the investment in behaviour modelling and data in one area is not easily re-useable due to the fundamental differences in implementation of that data in those systems. To address this issue, the UK Defence Science and Technology Laboratory (dstl) under the Centre for Defence Enterprise (CDE) set a challenge to investigate the feasibility of achieving methods for capturing models and data within common knowledge repositories for re-use across CGF and common modelling services. The purpose of this paper is to describe the work of Cranfield University at the Defence Academy to evaluate a new approach for knowledge capture and simulation agnostic execution of CGF Behaviours. It discusses the elements required to define a framework for the UK to develop a Behaviour Repository in the context of informing a possible future MoD CGF Service. This includes a discussion on the software architecture and processes; the lessons learned in the development of a multi-simulation behaviour authoring console; the skills required to populate such a repository and the governance required to provide a MoD CGF Service (including Non-run-time services). It describes the challenges both technical and non-technical in developing such a framework that enables greater re-use of verified and validated data models and behaviours across different simulation systems without being tied to a single CGF supplier.