Tool support for technical debt management: challenges, requirements and solutions
Technical debt (TD) describes the long-term consequences of technical decisions that make software easier to deliver in the short term but harder to maintain and evolve over time. Although tool support can reduce the effort required to manage TD, its practical value depends not only on technical capabilities, but also on how tools fit development workflows, support practitioners' decision-making, and match their organizational context.
This thesis investigates tool support for Technical Debt Management (TDM) from this broader perspective. First, we examined the state of the art of TDM automation and identified 121 automation artifacts, revealing an uneven distribution of support across TDM activities and TD types. We then investigated practitioners' perspectives, showing that they expect tools to fit existing workflows, provide useful information, and preserve human control. Further studies of practitioner discussions and CI/CD pipelines revealed challenges related to reliability, configuration, integration, feedback, and maintenance.
Building on these findings, we proposed and evaluated two solutions. TagDebt is a GitHub bot that integrates the identification of self-admitted technical debt into issue-tracking workflows. ATLAS-TD is a process that guides practitioners in selecting and adopting TDM tools according to their problems, requirements, workflows, and context.
Overall, this thesis shows that effective TDM tool support is a socio-technical and context-dependent concern. Effective tooling requires aligning technical capabilities with practitioner needs, development workflows, and organizational conditions.