The purpose of this study is to examine the current status of research software identified in STEM (Science, Technology, Engineering, Mathematics) repositories, specifically focusing on sharing, citations, citations based on sharing, authors, unique identifiers, and software types. This study examines both individual disciplines and the STEM field as a whole. The specific disciplines analyzed include astrophysics, biological sciences, computing, earth sciences, mathematics, and technology. The analysis was conducted on over 565,000 research software records indexed in the Data Citation Index (DCI), which indexes and tracks data from 453 repositories worldwide. The findings reveal that most STEM disciplines designate software types simply as “software,” and these records tended to have high citation rates. Exceptionally, the fields of computing and technology provided more detailed specifications for software types. In the computing field, various software types received citations relatively evenly; however, in the technology field, those simply labeled as “software” accounted for the majority of citations. In some instances across STEM fields, the software type was left blank. Regarding distribution, CRAN and Figshare were identified as the top two repositories for software sharing in STEM, whereas CRAN and Bioconductor emerged as the top two repositories for software citations. While most research software provided essential metadata, some critical information—such as author name and publication year—was occasionally missing. The majority of software listed individual researchers as co-authors rather than institutional authors. The contribution of this study lies in its comprehensive and detailed examination of the formal practices of research software across more than 453 global repositories, offering insights into both specific academic disciplines and the STEM field at large.