Search papers, labs, and topics across Lattice.
This study investigates the evolution of Agent Context Files (ACFs) in autonomous coding agents, focusing on how these files are maintained and their impact on code quality. By classifying ACF changes through a taxonomy rooted in software maintenance theory, the research reveals significant associations between specific types of changes and code quality outcomes. The findings highlight temporal patterns in ACF evolution throughout the development lifecycle, providing insights into effective governance of AI-driven software development processes.
ACF changes significantly influence code quality, revealing critical insights into how developers can better govern autonomous coding agents.
Context. Autonomous coding agents are increasingly used in software development, shifting parts of the engineering process to AI assistance. While this automation brings clear benefits, it introduces challenges in governance, traceability, and control over agent behavior. Agent Context Files (ACFs) have emerged as a practical mechanism to guide agents through structured instructions, yet little is known about how these artifacts are maintained and how their evolution relates to code development. Objective. This paper plans to investigate the evolution of ACFs and their role in agent-driven development. Specifically, we (1) classify ACF changes through a taxonomy grounded in software maintenance theory, (2) analyze how different types of changes are associated with code quality outcomes, and (3) examine their temporal patterns across the development lifecycle. Method. We conduct a large-scale mining study combining repositories with ACFs and agent-generated commits. We reconstruct ACF evolution at the commit level, classify changes using a qualitative approach, and analyze their association with code quality metrics. Statistical analyses and hypotheses are used to evaluate differences across maintenance categories, to inform future design of ACFs for governing autonomous coding agents.
