Analyzing and Mitigating Cost-Inefficient Behaviors in Coding Agents
Yiran Hu, Nan Jiang, Lin Tan and colleagues at Purdue University analyze 1,200 SWE-bench Verified trajectories from Claude Code and Mini-SWE-Agent to identify recurring behaviors that waste money in coding agents, then test three fixes over 10k more trajectories.
Ask this paper
Three wasteful behaviors. Subsumed retrieval (re-reading code already retrieved, often by a subagent), similar script generation, and repeated test execution. Together they occur in 79% to 98% of tasks and account for up to 22.75% of task cost.
Most common. Subsumed retrieval affects up to 92.33% of tasks and 11.41% of cost. Similar script generation occurs 5.91 to 9.98 times more often in Mini-SWE-Agent than in Claude Code.
Structure-aware retrieval is unreliable. It adds retrieval overhead and changes how the agent delegates, raising cost by up to 28.14% in some configurations.
Skills compared. Agent-synthesized skills produce low-level, trace-specific guidance and generalize poorly. Developer-designed skills with high-level guidance cut cost by up to 41.73%, roughly twice the best agent-synthesized result.
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.