A Study of the Reliability of Agentic AI-Generated Programs

arXiv 2609.18298 ↗First posted 16 Sep 2026cs.SE cs.AIcited by 0

On the map: LLM agents and tool use → Software engineering coding agents

Abstract (arXiv)

Agentic-AI based software development offers the promise of faster completion of the software, greater programmer efficiency, and more reliable code. The question is how can we verify these claims in an objective way? In this project, we attempted to answer this question based on three practices. First, we applied a typical best-practices agentic AI workflow for software development. Second, our target programs were ten well-known, release-quality human-written Linux utility programs so that we could compare the AI-generated code against a concrete ground truth. Third, we based our measure of reliability on a widely used testing technique, fuzz random testing. For this testing, we used both classic black box, generational testing and more modern coverage guided (gray box, mutational) testing using AFL++. We found that the AI-generated versions of the utility programs were typically as reliable - often more reliable - than the latest human-generated versions of these programs. While the AI-generated versions did have some failures, they were less common than the code from the standard repositories. Interestingly, the AI-generated code was less likely to have failures such as memory errors (such as buffer overflows) but more likely to have hangs such as infinite loops. In addition, we verified that generating robust and reliable software using agentic AI requires careful practice and human supervision. The quality of the code is highly dependent on the prompts and skills used, and how the human directing the process responds. We also demonstrated that using agentic AI workflow for software development (with its prompts and skills) can become a specification of the code that leads to cost-effective sustainability of the software.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use. Builds on 26 of its 57 references that are in the map. First posted on arXiv 16 Sep 2026.

Builds on

26 of 57 references.

Closest ancestors · 3

Same direction, newest first

Other references · 23

2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
2023MetaGPT: Meta Programming for A Multi-Agent Collaborative Frameworkcited by 1,808
2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,784
2022Large Language Models Are Human-Level Prompt Engineerscited by 1,045
2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 1,017
2023Teaching Large Language Models to Self-Debugcited by 988
2023Large Language Models Can Be Easily Distracted by Irrelevant Contextcited by 841
2022Decomposed Prompting: A Modular Approach for Solving Complex Taskscited by 558
2021Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributionscited by 431
2023Better Zero-Shot Reasoning with Role-Play Promptingcited by 223

11 more

2022Do Users Write More Insecure Code with AI Assistants?cited by 192
2025Checklists Are Better Than Reward Models For Aligning Language Modelscited by 128
2023Self-planning Code Generation with Large Language Modelscited by 124
2026Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forwardcited by 113
2022Capturing Failures of Large Language Models via Human Cognitive Biasescited by 98
2025RocketEval: Efficient Automated LLM Evaluation via Grading Checklistcited by 39
2024Prompting Techniques for Secure Code Generation: A Systematic Investigationcited by 34
2025The Lighthouse of Language: Enhancing LLM Agents via Critique-Guided Improvementcited by 20
2025SecureVibeBench: Benchmarking Secure Vibe Coding of AI Agents via Reconstructing Vulnerability-Introducing Scenarioscited by 13
2025Security Degradation in Iterative AI Code Generation -- A Systematic Analysis of the Paradoxcited by 10
2020The Relevance of Classic Fuzz Testing: Have We Solved This One?cited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way