Agyn: A Multi-Agent System for Team-Based Autonomous Software Engineering

TL;DR

This paper introduces Agyn, a multi-agent system that models software engineering as a team-based organizational process, achieving high task resolution rates without human intervention and outperforming single-agent baselines.

Contribution

The work presents a fully automated, multi-agent framework that replicates team roles, structured communication, and methodologies for autonomous software engineering.

Findings

Resolves 72.2% of tasks on SWE-bench 500

Outperforms single-agent baselines with similar models

Demonstrates the effectiveness of organizational design in autonomous systems

Abstract

Large language models have demonstrated strong capabilities in individual software engineering tasks, yet most autonomous systems still treat issue resolution as a monolithic or pipeline-based process. In contrast, real-world software development is organized as a collaborative activity carried out by teams following shared methodologies, with clear role separation, communication, and review. In this work, we present a fully automated multi-agent system that explicitly models software engineering as an organizational process, replicating the structure of an engineering team. Built on top of agyn, an open-source platform for configuring agent teams, our system assigns specialized agents to roles such as coordination, research, implementation, and review, provides them with isolated sandboxes for experimentation, and enables structured communication. The system follows a defined development methodology for working on issues, including analysis, task specification, pull request creation, and iterative review, and operates without any human intervention. Importantly, the system was designed for real production use and was not tuned for SWE-bench. When evaluated post hoc on SWE-bench 500, it resolves 72.2% of tasks, outperforming single-agent baselines using comparable language models. Our results suggest that replicating team structure, methodology, and communication is a powerful paradigm for autonomous software engineering, and that future progress may depend as much on organizational design and agent infrastructure as on model improvements.