Continuous Integration Practices (CIP) (\ie the socio-technical activities, and processes that govern how software changes are frequently integrated, built, and tested), are central to modern software development. While CIP provide rapid feedback and promote seamless collaboration, their growing scale and complexity introduce challenges in workflow maintenance, failure diagnosis, and security risk. These challenges can erode developers’ confidence, denoting that trust in CIP depends not only on technical soundness but also on the ability to comprehend and control the processes. We refer to this overarching construct as the \textit{trustworthiness of CIP}, which captures the extent to which developers can place justified confidence in automation despite its challenges and limitations.

In this thesis, we examine the \textit{trustworthiness of CIP} as a means to explain the dynamics of confidence in these practices. We argue that overlooking trustworthiness concerns undermines developers’ confidence and limits the effectiveness of CIP. We aim to quantify and characterize manifestations of untrustworthiness and to propose solutions that help developers sustain confidence in their CIP.