Dataset from "Unveiling Invisible Threats: An Empirical Study on the Prevalence and Co-occurrence of Configuration Smells in GitHub Actions Workflows" paper.

This dataset contains a collection of GitHub Actions workflows publicly available, annotated with configuration smells detected using the GASH (GitHub Actions Smell Hunter) tool. The data was collected to investigate the prevalence and co-occurrence of configuration smells in CI/CD pipelines, which can potentially compromise the reliability, security, and maintainability of software development workflows. The dataset consists of 3.996 repositories, each containing 16.572 configuration files, which were analyzed for 16.882 different types of configuration smells. The dataset includes metadata about each workflow, such as the repository owner, repository name, and number of stars, as well as the results of the smell detection analysis, including the type and frequency of each smell. This dataset was used in a research study to explore the prevalence and co-occurrence of configuration smells in GitHub Actions workflows, and to identify patterns and trends in configuration practices that can inform the development of better practices and guidelines for CI/CD pipeline management.