Gait recognition has appeared as an important biometric strategy because of its non-intrusive attributes and straightforward implementation, facilitating identification without physical contact. In contrast to systems reliant on silhouette information and other visual attributes, skeleton-based approaches retrieve gait data independently of appearance indicators. Nevertheless, conventional methods in this field generally depend on manually prepared features and adjacency matrices that are exclusively dependent on the physical connectivity of joints. This dependence is a significant obstacle in obtaining semantically rich representations of the joint interactions and fundamental motion patterns essential for practical gait analysis. This paper introduces a skeleton-based Multi-Part Attention-Guided (MPA) Spatial-temporal Graph Convolutional Networks (ST-GCNs) gait recognition approach, MAST-GCN, which enhances the modeling of spatial and temporal dependencies in skeletal data through a multi-part attention mechanism. Unlike ST-GCNs, which depend on rigid graph structures and struggle to capture long-range interactions essential for identifying subtle gait differences, our method divides the skeleton into distinct anatomical regions and applies a Part-wise Attention module. By integrating attention-weighted features through a hierarchical fusion process, the model effectively captures both detailed and broad gait patterns across multiple temporal scales. Tested on benchmark datasets like CASIA-B and OUMVLP-Pose, attaining rank-1 precisions of 95.9%, 91.8%, and 88.6% under normal walking (NM), carrying bag (BG), and wearing coat (CL) conditions, respectively, on CASIA-B dataset and 91.7% on the OUMVLP-Pose dataset, showing superior performance. Our approach performs better than state-of-the-art methods, particularly highlighting the benefits of part-based, attention-driven feature extraction for robust and precise gait recognition.