Faster Code, Deeper Debt? A Multivocal Literature Review on Technical Debt and Its Early Signs in LLM-Assisted Software Development

Collection of recent SE/AI papers highlighting security-relevant risks and mitigations when integrating LLMs into software development. Key findings: LLM-assisted coding amplifies traditional technical debt (code, design, docs) and introduces new LLM-specific debts (fast-integration, prompt, data, provenance, ethical, governance) that increase maintenance and security exposure; there are no standard LLM-specific metrics or benchmarks yet. Several papers propose mitigations with direct security relevance: human-in-the-loop controls, prompt engineering, provenance and data-quality alignment, and