The framework of defect phase diagrams is a powerful, robust, knowledge-based design strategy that can be used to manipulate defect phases using defect-property relationships for processing tailored materials. To this end, chemical potential is used as the key variable to develop automatic workflows for calculating defect phase diagrams. Following results from high-resolution transition electron microscopy, binary defect phase diagrams have been generated for $\Sigma$7 [0001] 21.78$^{\circ}$ Mg grain boundaries with Ga and Ca. The stability of the two known defect phases, A- and T-type grain boundary, and the chemical trends for the relative stability of different atomic configurations are derived using ab-initio simulations. Stabilising the A-type, Ga segregates to the grain boundary beyond the dilute limit and induces a systematic transition of the preferred segregation sites. However, Ca is observed to stabilise the T-type grain boundary and segregate to the grain boundary within the dilute limit. Connecting the workflow with experimental datasets, a good agreement has been confirmed. Extending to other planar defects, the influence on the deformation behaviour is then discussed. The advantages of automized workflows for the defect phase diagrams are then demonstrated to extend the considerations to finite temperatures and free energies calculated within the quasi-harmonic approximation.