Analyzing and Mitigating Cost-Inefficient Behaviors in Coding Agents

2026-09-24

AI Summary

This study addresses the underexplored behavioral inefficiencies of coding agents, which incur substantial monetary costs. By systematically analyzing 1,200 execution traces on the SWE-bench benchmark, this work identifies three categories of inefficient behaviors affecting 79%–98% of tasks and evaluates mitigation strategies, including structure-aware retrieval and skill mechanisms, revealing potential side effects of the former. Furthermore, it demonstrates that developer-designed skills outperform automatically synthesized ones in generalization capability. Applying such curated skills reduces agent execution costs by up to 41.73%, achieving approximately twice the effectiveness of automatic synthesis. These findings provide empirical evidence for developing cost-efficient coding agents.

Abstract

Although effective, coding agents often incur substantial monetary costs. Their recurring cost-inefficient behaviors remain underexplored. We conduct the first study of behavioral cost inefficiencies in coding agents, analyzing 1,200 trajectories from Claude Code and Mini-SWE-Agent across four configurations on SWE-bench Verified. We identify three cost-inefficient behaviors: subsumed retrieval, similar script generation, and test re-execution. We then evaluate three mitigation strategies: structure-aware retrieval, agent-synthesized skills, and developer-designed skills, over 10k trajectories on held-out SWE-bench Verified and Pro tasks. Our main findings are: (1) The three behaviors affect 79.00\%--98.00\% of coding tasks and account for up to 22.75\% of task cost. (2) Structure-aware retrieval can introduce retrieval overhead and alter agent delegation, causing inconsistent improvements in retrieval efficiency and cost increases of up to 28.14\%. (3) Agent-synthesized skills tend to produce low-level, trace-specific guidance, limiting their effectiveness and generality. (4) In contrast, developer-designed skills provide high-level, trace-agnostic guidance, reducing cost by up to 41.73\%, roughly twice the maximum gain from agent-synthesized skills.

Problem

• Coding Agents
• Cost-Inefficiency
• Behavioral Analysis
• Cost Mitigation
• LLM-based Agents

Innovation

• coding agents
• cost inefficiency
• trajectory analysis
• developer-designed skills
• mitigation strategies