RepoInsights: Leveraging GitHub Traces for Software Process Metrics

Software process metrics often rely on declarative data sources such as self-reported progress at daily meetings or manual inputs in project management tools. These approaches introduce subjectivity and limited traceability. In contrast, GitHub repositories contain rich, automatically generated traces offering more concrete and reproducible data for empirical analysis that can mitigate subjectivity. This paper introduces RepoInsights, a system designed to extract, store, and visualize key repository-based software process metrics derived from GitHub traces. We begin with a systematic literature review to compile existing process metrics, which are then operationalized using elements such as commits, pull requests, and issues as proxies. The metrics include, among others, Bus Factor, Time to First Response, and Pull Request Closure Rate. The tool provides dashboards tailored for project managers and researchers, enabling data-driven decision-making. The system’s effectiveness was validated through structured user testing involving two target profiles: software researchers and project managers. Results demonstrate the tool’s utility in teams’ assessment, comparison, and monitoring, offering a replicable, trace-based alternative to traditional metric gathering in software process improvement efforts.