Trust but Verify? Uncovering the Security Debt of Autonomous Coding Agents

2026-07-14

AI Summary

This study systematically investigates the security debt introduced by autonomous coding agents, which, while enhancing development efficiency, generate high-risk vulnerabilities often missed by conventional human code reviews. Leveraging the AIDev dataset, a validated LLM-as-a-judge evaluation framework, and qualitative human analysis, the authors find that 38.9% of agent-generated pull requests contain security smells, with 82.3% pertaining to supply chain integrity and 99.6% of critical smells involving hardcoded credentials. Notably, 81.1% of these credential leaks originate from human collaborators and evade detection by existing review mechanisms, exposing a critical blind spot in current development workflows.

Abstract

The increasing adoption of autonomous coding agents accelerates software development but also introduces scoped security risks within high-impact file paths that can outpace traditional human review capacity. While prior research has primarily evaluated these systems in terms of functional correctness and productivity, this paper presents a large-scale empirical study using the AIDev dataset to systematically characterize security code smells in agent-generated pull requests (PRs). Through a combination of a validated LLM-as-a-judge framework and manual qualitative analysis, we identify and classify security misconfigurations across 16,112 file changes spanning 4,022 pull requests. Our results reveal that 38.9% of agent-generated PRs contain at least one security smell, with supply chain integrity issues accounting for 82.3% of all detected security smells. Furthermore, hard-coded credentials constitute 99.6% of all critical-severity security smells. Crucially, we find that human collaborators are responsible for introducing 67.6% of genuine leaked secrets within these agent-assisted workflows, while existing automated and human review processes fail to detect 81.1% of these credentials prior to integration. These findings highlight substantial security risks in agent-assisted software development workflows and suggest a potential reduction in developer vigilance. They also underscore the urgent need for context-aware security guardrails implemented directly at the point of human-AI collaboration.

Problem

• autonomous coding agents
• security debt
• security code smells
• hard-coded credentials
• supply chain integrity

Innovation

• autonomous coding agents
• security code smells
• LLM-as-a-judge
• hard-coded credentials
• supply chain integrity