Wink: Recovering from Misbehaviors in Coding Agents

TL;DR

This paper introduces Wink, a system that automatically detects and corrects misbehaviors in autonomous coding agents powered by large language models, significantly improving their reliability and reducing manual interventions.

Contribution

We propose a taxonomy of agent misbehaviors and develop Wink, a lightweight, asynchronous self-intervention system that effectively recovers from these issues at scale.

Findings

Wink resolves 90% of single-intervention misbehaviors

Reduces Tool Call Failures, Tokens per Session, and Engineer Interventions

Successfully deployed and tested on over 10,000 real-world trajectories

Abstract

Autonomous coding agents, powered by large language models (LLMs), are increasingly being adopted in the software industry to automate complex engineering tasks. However, these agents are prone to a wide range of misbehaviors, such as deviating from the user's instructions, getting stuck in repetitive loops, or failing to use tools correctly. These failures disrupt the development workflow and often require resource-intensive manual intervention. In this paper, we present a system for automatically recovering from agentic misbehaviors at scale. We first introduce a taxonomy of misbehaviors grounded in an analysis of production traffic, identifying three primary categories: Specification Drift, Reasoning Problems, and Tool Call Failures, which we find occur in about 30% of all agent trajectories. To address these issues, we developed a lightweight, asynchronous self-intervention system named Wink. Wink observes agent trajectories and provides targeted course-correction guidance to nudge the agent back to a productive path. We evaluated our system on over 10,000 real world agent trajectories and found that it successfully resolves 90% of the misbehaviors that require a single intervention. Furthermore, a live A/B test in our production environment demonstrated that our system leads to a statistically significant reduction in Tool Call Failures, Tokens per Session and Engineer Interventions per Session. We present our experience designing and deploying this system, offering insights into the challenges of building resilient agentic systems at scale.