The authors present the first step in the development of new CGF (Computer Generated Forces) to provide CIMIC (Civil Military Co-operation) Modeling capabilities in the complex and critical sector of military operation planning training. The proposed CGF, based on intelligent software agents, were demonstrated, in terms of feasibility and initial development, by PIOVRA (Poly-Functional Intelligent Operational Virtual Reality Agents) research project results. These new CGF are designed as HLA federates so that external federated simulation systems can simulate the effects of CIMIC activities more realistically and in greater detail: i.e. the perception of the presence of different ethnic civilian groups on their territory. This supports training activities for planners or Operations Commanders, providing realistic scenarios in which they must choose between different options with a higher degree of certainty that the foreseen results will be similar to what is expected from their actual implementation. The authors are proposing a general architecture and agent structure for demonstrating and testing these concepts by integrating these CGF in an HLA federation with a constructive simulator. As a result, it will be possible to evaluate the impact of this approach when combined with existing systems and to carry out VV&A (Verification, Validation and Accreditation) on these developments.