Type: url
Original: https://arxiv.org/pdf/2606.14796
Status: generated
Summary
This multivocal literature review examines 104 sources, 31 formal and 73 grey, on how LLM-assisted development contributes to technical debt and what strategies, tools, metrics and benchmarks exist to manage it. It finds that LLMs amplify traditional code, design and documentation debt while introducing LLM-specific categories, notably fast-integration debt, which triggers a domino effect leading to governance debt, alongside prompt, ethical, data and provenance debt. It reports human-in-the-loop frameworks, prompt engineering and data-quality alignment as the main mitigation strategies, SonarQube as the most commonly used detection tool, and no standardised benchmarks or LLM-specific metrics yet in existence.