The rapid expansion of the digital economy is reshaping urban systems, yet the pathways through which digitalization drives urban green land use transition remain insufficiently understood. Using panel data from 279 Chinese cities, we measure Urban Green Land Use Efficiency (UGLUE) via a super-efficiency Slack-Based Measure (SBM) model and estimate digital economy effects through double machine learning (DML). We find that digital technology, digital industry, and digital infrastructure all positively influence UGLUE, with digital technology exerting the strongest effect, followed by digital industry, while digital infrastructure exerts the weakest direct effect. Urban spatial compactness (USC) mediates this relationship, functioning as a dominant transmission channel for both digital industry and digital technology, and a supplementary yet significant pathway for digital infrastructure, indicating that digitalization enhances UGLUE in part by promoting more compact urban forms. Effects are heterogeneous, as resource-based and old industrial cities benefit more from technological upgrades, while cities with higher administrative status gain more from broader digital development. These findings identify USC as a key transition pathway linking digitalization to sustainable land use outcomes, and provide evidence-based support for governance-differentiated digital economy policies that steer urban land use transition toward green and compact development trajectories.