Kozuchi Agent: A Language-Agnostic Open-Weight Agent for Software Repair

2026-08-16

AI Summary

This study addresses the challenges of prolonged development cycles, tool dependency, and cross-language evaluation in industrial-grade software repair by proposing a language-agnostic open-source agent architecture alongside a reusable CI evaluation pipeline. Through explicit stage planning, persistent state management, and a cross-agent test-time selection mechanism, the framework achieves an auditable and deterministic repair process. Leveraging the Qwen3.5-27B model, this method attains a 74.8% resolution rate on SWE-bench Verified, ranking first among open-source solutions for both Python and Java. Furthermore, it reduces operational touchpoints by 80%, significantly enhancing the efficacy and reproducibility of open-source models in complex engineering tasks.

Abstract

Industrial software-engineering teams increasingly need LLM agents that turn bug reports into correct patches, yet benchmark-scale operation adds long horizons, tool-use discipline, context persistence, heterogeneous clusters, and evaluation reuse. We present Kozuchi Agent, a language-agnostic open-weight repair agent and CI-operated evaluation pipeline. Explicit phases, persistent state, deterministic tools, a model-independent action interface, and cross-agent test-time selection make runs auditable and repeatable. With locally hosted Qwen3.5-27B, no fine-tuning, and TTS@8, Kozuchi resolves 374/500 SWE-bench Verified instances on the official evaluator. Unchanged on Multi-SWE-bench Java, the same 27-billion-parameter agent resolves 41/128 instances (32.03%), ranking first among strict open-weight submissions and fourth of 42 overall; on Python it ranks 12th of 135 and first among open-weight systems. Per-phase behavior remains within +/-5 percentage points across languages. Remaining failures mainly reflect semantic correctness, Java-specific harness issues, and selection errors. Across both tracks, results compare favorably with open/local peers by parameter count. Analysis of candidate diversity, selector regret, and patch reliability shows that the remaining gap is primarily semantic correctness and selection rather than edit formatting or proprietary-model access. Operationally, reusable CI stages reduce operator touch-points from five to one across heterogeneous internal clusters.

Problem

• Software Repair
• LLM Agents
• Language-Agnostic
• SWE-bench
• Semantic Correctness

Innovation

• Language-Agnostic Agent
• Open-Weight Model
• Test-Time Selection
• CI-Operated Evaluation
• Deterministic Tool Use