Agent-based simulation models provide a viable approach for developing applied models of species and systems for predictive management. However, there has been some reluctance to use these models for policy applications due to complexity and the need for improved testing and communication of the models. We present the development and testing of a comprehensive model for Skylark (Alauda arvensis) in Danish agricultural landscapes. The model is part of the ALMaSS system, which considers not only individual skylarks, but also the detailed dynamic environment from which they obtain the information necessary to simulate their behaviour. Population responses emerge from individuals interacting with each other and the environment. Model testing was carried out using pattern-oriented modelling. The testing procedure was based on the model’s ability to represent detailed real world patterns of distribution and density, reproductive performance and seasonal changes in territory numbers. Data to support this was collected over a 13-year period and comprised detailed field observations of breeding birds and intensive surveys. The model was able to recreate the real world data patterns accurately; it was also able to simultaneously fit a number of other secondary system properties which were not formally a part of the testing procedure. The correspondence of model output to real world data and sensitivity analysis are presented and discussed, and the model’s description is provided in ODdox format (a formal description inter-linked to the program code). Detailed and stringent tests for model performance were carried out, and standardised model description and open access to the source code were provided to open development of the skylark model to others. Over and above documenting the utility of the model, this open process is essential to engender the user trust and ensure continued development of these comprehensive systems for applied purposes.