Digital neuropsychology uses computational tools to improve traditional assessments, such as the Trail-Making Test (TMT). While the paper-and-pencil TMT primarily measures completion time, digital adaptations enable the collection of continuous behavioural data. These metrics offer new markers of cognitive decline. Following this path, we developed a computerised TMT (cTMT) preserving the original structure while recording high-resolution mouse trajectories across several trials. Seventy-four older adults (41 with mild cognitive impairment and 33 controls) completed the cTMT, as well as a standard diagnostic battery. We extracted several features from the cursor time series, including reaction times, speed and acceleration metrics, trajectory deviations and state-based measures. The results showed that demographic models provided only modest discrimination (AUC = 0.56). Digital hand features improved performance (AUC = 0.65), and combining them with demographics reached an AUC of 0.71. Which is closer to the ceil performance (AUC = 0.73) achieved by the neuropsychological battery that was the quantitative part of the criteria to define the classes in the first place. These findings show that digital biomarkers can achieve performance comparable to standardized tests. This cost-effective cTMT, compatible with hardware available at home or at clinical facilities, will be further validated with multimodal data for the early detection and monitoring of cognitive decline.