CodeRescue: Budget-Calibrated Recovery Routing for Coding Agents
2026-07-21 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors study how coding agents decide between trying more cheap, quick recovery steps or escalating to a costlier, stronger model after a failed attempt. They create a system that learns to choose the best action based on past experience and add a method to adjust decisions based on budget constraints without retraining. Tested on five coding tasks, their approach balances fixing mistakes cheaply and escalating effectively, improving success while saving resources compared to simpler strategies. The system is publicly available for others to use.
coding agentsexecution feedbackcost-aware systemscascade modelsbudgeted deploymentsupervised routingConformal Risk Control (CRC)exchangeabilityrecovery routingbenchmark evaluation
Authors
Qijia He, Jiayi Cheng, Chenqian Le, Rui Wang, Xunmei Liu, Yixian Chen, Jie Mei, Zhihao Wang, Xupeng Chen, Yuhuan Chen, Tao Wang
Abstract
Coding agents increasingly operate in executable environments where a failed attempt produces actionable feedback rather than merely an incorrect answer. Existing cost-aware systems typically treat such failures as cascade decisions: try a cheap model first, then escalate hard cases to a stronger and more expensive model. In coding, however, execution feedback can also make further cheap-model recovery worthwhile, raising a budgeted deployment question: when should an agent spend more cheap compute, and when should it escalate? We formulate this post-failure decision as recovery routing over heterogeneous actions and train a supervised router from execution rollouts. To make the same router usable under changing budgets, we add a Conformal Risk Control (CRC) layer that selects a deployment-time cost penalty without retraining and provides marginal expected-cost control under exchangeability. Across held-out failures from five coding benchmarks, cheap recovery and escalation exhibit complementary success patterns. The calibrated frontier improves over fixed actions, prompt-only routers, and a binary cascade baseline; in the main GPT-5.4-nano/GPT-5.4 setting, one CRC-calibrated frontier point exceeds always-escalate solve rate while using 35% of its mean recovery cost. Code is available at https://github.com/Qijia-He/agent-budget-control.