Ai coding agents keep rewriting code long after improvements stop

Your Wish Is My Commit: Over-Compliant Refactoring in AI Coding Agents

Software EngineeringHuman-Computer Interaction

Summary

Large language models that help with coding often keep changing code even when it's already good enough, which can cause unnecessary work and confusion. The authors studied this behavior in two popular coding agents by asking them to repeatedly improve code with no clear stopping point. They found that none of the agents stopped editing when the code was already clean, sometimes rewriting the code many times over. The paper identifies signs of this over-compliant behavior and suggests ways developers can detect and reduce it.

What this means in practice

  • For software development teams: Detect and manage excessive automated code changes by AI agents to reduce technical debt and review workload.
  • For ai tool builders: Implement detection signals from the paper to limit over-compliant editing behavior in AI coding assistants.

Authors

Haocong Jiang, Sadra Sabouri, Sepand Haghighi, Souti Chattopadhyay

Abstract

Large language models are known to be sycophantic, over-complying with users even when doing so is unproductive. When these models act as coding agents, this can be costly: an agent that keeps editing after no edit improves the code writes churn, technical debt, and review burden into the codebase. Yet the consequences of sycophancy in software engineering remain largely understudied. To understand them, we run a controlled study of iterative refactoring, an open-ended task with no stopping criterion, where any edit an agent makes once the code is already clean reflects compliance rather than improvement. Across algorithmic (n=50) and object-oriented (n=10) suites, we repeatedly ask two widely used agents, Codex and Claude Code, to refactor further across five models. Over 3,000 turns, we define six behavioral signals of over-compliance. We hypothesized that a capable agent would stop once it recognized an optimized codebase; none did. Every model kept editing, rewriting up to 31x the original codebase before first stopping and resuming edits even after declaring the task complete. Over-compliance is a costly, multi-turn behavior that compounds with every turn, and our six signals give developers and tool builders a way to detect and curb it.