Systematic Literature Review of Explainable LLM-Powered ChatOps for CI/CD Pipeline Diagnostics and Developer Support

Abstract

This systematic literature review examines how explainable Large Language Models (LLMs) transform chat-ops applications for CI/CD pipeline diagnostics and developer support. Analyzing contemporary research from 2021–2025, this review synthesizes findings from studies on conversational AI agents, LLM-driven DevOps automation, and explainable AI techniques in software development. Key methodological approaches include Domain-Specific Languages (DSLs) for chatbot development, multi-agent orchestration frameworks, retrieval-augmented generation (RAG) systems, and two-staged LLM frameworks for detecting and remediating failures. Key findings reveal significant advancements in automating root cause analysis, orchestrating pipelines intelligently, and creating natural language interfaces for complex DevOps operations. Integrating explainable AI principles with conversational interfaces improves developer productivity, reduces mean time to resolution (MTTR), and enhances collaboration between development and operations teams. However, persistent interpretability, security governance, integration complexity, and maintaining human oversight in automated decision-making processes. This review provides a comprehensive roadmap for researchers and practitioners seeking to leverage explainable LLM-powered chat-ops solutions for enhanced CI/CD pipeline management and developer support.

Ähnliche Arbeiten