Understanding community structure in complex social networks is both challenging and essential. We present two novel algorithms—CNM-ES and RECC-SC—that integrate size constraints into classic community detection frameworks, ensuring the discovery of robust and interpretable clusters. CNM-ES refines traditional agglomerative methods by halting merges that would compromise community integrity, while RECC-SC augments a divisive approach with a minimum size parameter to prevent trivial partitions. Evaluations on synthetic benchmarks and real-world DBLP collaboration networks demonstrate that our methods consistently uncover meaningful communities that honor user-defined size limits. We also provide an user-friendly web application that enables interactive exploration and analysis of detected communities.