A Study of Library Usage in Agent-Authored Pull Requests

TL;DR

This study analyzes how AI coding agents use libraries in pull requests, revealing frequent library imports, cautious dependency additions with versioning, and diverse library choices, providing insights into their interaction with software ecosystems.

Contribution

It provides the first large-scale empirical analysis of library usage patterns in agent-generated pull requests, highlighting differences from traditional LLM behaviors.

Findings

Agents import libraries in 29.5% of PRs

Only 1.3% of PRs add new dependencies

75% of dependency additions specify a version

Abstract

Coding agents are becoming increasingly capable of completing end-to-end software engineering workflows that previously required a human developer, including raising pull requests (PRs) to propose their changes. However, we still know little about how these agents use libraries when generating code, a core part of real-world software development. To fill this gap, we study 26,760 agent-authored PRs from the AIDev dataset to examine three questions: how often do agents import libraries, how often do they introduce new dependencies (and with what versioning), and which specific libraries do they choose? We find that agents often import libraries (29.5% of PRs) but rarely add new dependencies (1.3% of PRs); and when they do, they follow strong versioning practices (75.0% specify a version), an improvement on direct LLM usage where versions are rarely mentioned. Generally, agents draw from a surprisingly diverse set of external libraries, contrasting with the limited "library preferences" seen in prior non-agentic LLM studies. Our findings offer an early empirical view on how AI coding agents interact with today's software ecosystems.