Newest papers

PAUSE: Search-Free Steering of Coding Agents Before Their First Source-Code Edit

Viral Patel, 谭博引, Jason Xue, Youcheng Sun · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29870 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

Los métodos de control en línea existentes revisan las trayectorias periódicamente o invocan un asesor de verificación cuando un monitor detecta problemas de comportamiento; las revisiones frecuentes son costosas, y la ausencia de comportamiento anómalo no significa que un agente esté en el camino correcto. Proponemos PAUSE, un marco para el control de los agentes de codificación con una sola intervención de planificación. Un agente de verificación logra su primer código fuente editado a una media del 60% del progreso de la trayectoria, después de la exploración, la reproducción de errores y las pruebas. PAUSE intercepta esta solicitud de edición antes de que se ejecute, lo que no requiere ningún monitor aprendido, ninguna búsqueda en puntos de control, y ninguna llamada de modelo adicional (de forma libre de búsqueda). Un

Machine translation (Meta M2M100); the English original follows.

Abstract

Coding agents can invest substantial effort in repository exploration and testing yet still produce unsuccessful patches. Existing online steering methods review trajectories periodically or invoke an advisor when a monitor detects behavioral problems; frequent reviews are costly, and the absence of anomalous behavior does not mean that an agent is on the right track. We propose PAUSE, a framework for steering coding agents with a single planning intervention. Agents reach their first source-code edit at a median of 60% of trajectory progress, after exploration, bug reproduction, and testing. PAUSE intercepts this edit request before it executes, which requires no learned monitor, no search over checkpoints, and no extra model call (search-free). A separate Planner reads the task and the pre-edit transcript and writes one repair plan; the same agent, the Executor, resumes with this plan, keeping its context and workspace. PAUSE applies online, during a live run, and offline, from a recorded trajectory prefix. We evaluate PAUSE on SWE-bench Verified and SWE-bench Multilingual with two Executors and three agent harnesses. With Qwen3-coder as Executor and GPT-5.6 Sol as Planner, online steering improves resolution by 16.8% on Verified in OpenHands and by 14.7% on Multilingual in SWE-agent. In SWE-agent, PAUSE outperforms LivePlan, a rule-triggered advisor, by 3.2% on Verified and 3.7% on Multilingual at similar Planner cost, and is comparable to SICA's periodic overseer at about 11-15% of its Planner cost. Offline, PAUSE recovers 35.4% of eligible failed Verified tasks, and resuming from the recorded prefix uses 33-41% fewer Executor tokens than restarting.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Executor, Computer science, Coding (social sciences), Planner, Context (archaeology), Encoder

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange