AI-Generated Code Is Not Reproducible (Yet): An Empirical Study of Dependency Gaps in LLM-Based Coding Agents

TL;DR

This empirical study examines the reproducibility of LLM-generated code, revealing that only about two-thirds of projects run successfully in clean environments and highlighting significant hidden dependencies.

Contribution

The paper introduces a dependency framework and provides the first large-scale empirical analysis of reproducibility issues in LLM-based coding agents.

Findings

68.3% of projects execute successfully out-of-the-box

Substantial variation in reproducibility across programming languages

Average expansion of dependencies by 13.5 times from declared to runtime

Abstract

The rise of Large Language Models (LLMs) as coding agents promises to accelerate software development, but their impact on generated code reproducibility remains largely unexplored. This paper presents an empirical study investigating whether LLM-generated code can be executed successfully in a clean environment with only OS packages and using only the dependencies that the model specifies. We evaluate three state-of-the-art LLM coding agents (Claude Code, OpenAI Codex, and Gemini) across 300 projects generated from 100 standardized prompts in Python, JavaScript, and Java. We introduce a three-layer dependency framework (distinguishing between claimed, working, and runtime dependencies) to quantify execution reproducibility. Our results show that only 68.3% of projects execute out-of-the-box, with substantial variation across languages (Python 89.2%, Java 44.0%). We also find a 13.5 times average expansion from declared to actual runtime dependencies, revealing significant hidden dependencies.