Иллюзия надежности: почему привычка копировать готовые скрипты автоматизации тихо разрушает безопасность софта
Automation is fundamental to modern software engineering, with platforms such as GitHub providing built-in systems like GitHub Actions to streamline tasks across the software development lifecycle. Similar to traditional open-source software reuse, automated workflows rely heavily on reusable components such as Actions and reusable workflows. While this reliance offers significant productivity benefits, it also introduces critical dependency-related risks within the software supply chain, particularly regarding security vulnerabilities and maintainability. To address these overlooked challenges, this PhD dissertation presents a comprehensive empirical investigation into automation and reuse practices within the GitHub Actions ecosystem. Employing a mixed-methods approach, it combines qualitative survey findings from workflow practitioners with large-scale software repository mining. Through this empirical lens, the research evaluates workflow reuse behaviours, quantifies component outdatedness, maps complex dependency networks, and assesses ecosystem-wide security dimensions. The findings reveal key motivations and selection criteria behind workflow reuse while uncovering widespread vulnerability exposure and maintenance challenges. Ultimately, this work provides actionable guidance and insights for practitioners, tool developers, and researchers to foster a more secure, reliable, and maintainable workflow automation ecosystem.
## В чем суть открытия Использование чужих готовых скриптов CI/CD и автоматизации напоминает подключение электроприборов через цепочку случайных переходников: при устаревании или уязвимости хотя бы одного звена вся инфраструктура проекта подвергается скрытой компрометации. ## Как это устроено: принцип работы Modern continuous integration relies on reusable GitHub Actions composed into workflow steps, which predominantly execute JavaScript routines or containerized environments. These Actions depend heavily on deeply nested npm dependency trees that concentrate around a handful of official packages, while workflow authors routinely pin only major versions or copy-paste code. As a result, unpatched upstream vulnerabilities and code-level sanitization flaws silently propagate downstream into the build environments of millions of software repositories. ## Результаты и доказательства - **Комплексный скор значимости:** 9.5 / 10.0 - **Уровень доказательности:** code_benchmark - **Бенчмарки и прирост:** Technical lag (months), opportunity lag (months), CVE vulnerability count, CodeQL CWE weakness density, Action update frequency, reuse proportion (%) - **Размер выборки / валидация:** N = 419 survey respondents; 22,758 repositories; 979,591 workflows; 7,927,295 steps; 8,107 JavaScript Actions across 47,906 releases and 2.2M+ dependencies - **Реализация в открытом доступе:** Да (верифицируемый репозиторий кода) ## Что это меняет на практике Provides engineering teams and DevOps architects with concrete empirical justification to implement automated dependency auditing specifically for workflow files, replace brittle copy-pasting with verified composite Actions, and enforce commit-hash pinning on critical deployment pipelines to prevent supply chain tampering. ## Границы метода и риски The empirical dataset focuses on public GitHub repositories with at least 100 stars and 100 commits, potentially under-representing private enterprise workflows, while deep dependency and vulnerability analyses are restricted to JavaScript Actions rather than Docker or composite configurations.