Closed-Loop Autonomous Software Development via Jira-Integrated Backlog Orchestration: A Case Study in Deterministic Control and Safety-Constrained Automation

A new research paper by Elias Calboreanu presents a groundbreaking case study in autonomous software development. The system is not a simple code generator but a full control architecture that orchestrates a software backlog of approximately 1,602 tasks. It integrates directly with Jira for status tracking and uses a complex automation stack of about 12,661 lines of Python code and 6,907 lines of versioned prompt specifications. The core innovation is its deterministic, seven-stage pipeline with built-in safety mechanisms like 101 exception handlers, 12 lock mechanisms, and structured context packages to bound AI assistance.

In formal evaluation, the system demonstrated remarkable reliability. Its initial 152-run window yielded a 100% terminal-state success rate, with a statistical confidence interval of [97.6%, 100%]. It has since accumulated over 795 run artifacts in continuous operation. The paper details rigorous testing, including three rounds of adversarial code review that identified 51 findings, all of which were closed. In a specific test on 10 security tickets, the system autonomously completed six, required manual help for two, and closed two by policy. The results prove that embedding AI autonomy within explicit control, recovery, and audit frameworks makes large-scale, traceable software automation a practical reality.