An Ai-Driven Devops Artefact For Automated Code Modification and Continuous Software Quality Assessment

Maintaining software quality across large, multi-repository codebases is a persistent challenge, especially when teams lack dedicated quality assurance resources. Although Large Language Models (LLMs) can generate code modifications that improve quality, existing approaches treat code generation, continuous integration, and quality monitoring as isolated concerns. This paper presents the design of an AI-driven DevOps artefact, developed following Design Science Research (DSR), that orchestrates the full cycle from LLM-based code modification through CI/CD validation to quality assessment and visualisation. Deployed on a commercial platform with 56 repositories and approximately 93,000 lines of code, the artefact reduced multi-repository modification effort to a single pipeline execution, enforced quality gates on every change, and provided the development team with continuous, crossrepository visibility into metrics such as code coverage, vulnerabilities, bugs, security hotspots, and duplications - with human intervention limited to prompt selection, script invocation, and pull request approval.