Abstract: The first step in improving a model is to identify deficiencies in the model forecast. With the continuous advancement of numerical prediction technology, the precise assessment and analysis of model prediction errors, particularly the traceable technology of systematic errors, has gradually become a pivotal issue in model evaluation. Atmospheric energy circulation, as a fundamental principle of atmospheric motion, accurately represents the dynamic and physical interaction mechanisms. With a deeper understanding of the atmospheric energy cycle process, its applications have also expanded continuously. Particularly in recent decades, it has been used to assess the performance of numerical models and reanalysis datasets, serving as an essential metric for understanding model forecast capability and identifying systematic errors. Encompassed within the mixed space-time domain energy cycle are the mean circulation, stationary (deviation from the zonal mean), and transient (deviation from the temporal mean) eddies, and their interconversions of the available potential energy and kinetic energy, with each component holding physical significance. Based on the mixed space-time domain energy cycle framework and scale analysis methods, the energy cycle error characteristics and sources in CMA-GFS at planetary scales (zonal wavenumber 1-3) and synoptic and below (greater than zonal wavenumber 3) scales are examined using CMA-GFS global forecast product and ERA5 global reanalysis data in 2022. Results show that CMA-GFS can effectively replicate characteristics of the atmospheric energy cycle. However, its overestimation of baroclinity results in a stronger available potential energy of the mean circulation, which shows an increasing trend with forecast lead time. The stationary and transient eddy energy are dominated by planetary scales and synoptic and below scales, respectively. Errors in the available potential energy of the stationary eddy component and transient eddy component are determined by thermal conditions. CMA-GFS shows higher stationary eddy available potential energy and less transient eddy available potential energy. Systematic underestimations are observed in kinetics of stationary eddy component and transient eddy component, with predominantly negative errors concentrated near centers of subtropical jets and the polar night jet. This is primarily due to stronger barotropic transports, which transfer more energy from eddies to the mean circulation. As the baroclinity gradually increased, the transient eddy also increased after 120 h lead time. CMA-GFS underestimates four eddy energies in the boreal winter and overestimates or slightly underestimates them in the boreal summer, leading to a significant weakening of their seasonal variation.