One of the popular ways to clutter reduction techniques is to combine neighboring points into one marker that somehow shows that it contains multiple entities - this way is called clustering. In this paper, we present a JavaScript library to define optimal size of clusters and render them. Moreover, markers have to present heterogeneous data inside of clusters. The presented library relies on server side clustering, no matter if is it a real-time clustering or a static bunch of hexagonal grids. For the library, a server provides the bunch of grid layers by different cell sizes - from smaller to larger. The library relies on data fetching provided by external library, such as Mapbox/Maplibre, so it can work with both GeoJSON and vector tiles. Using the HTML Canvas to render the marker allows to full customizing the marker image: manage the colors and proportions of cluster fractions and the size.