PentestEval: Benchmarking LLM-based Penetration Testing with Modular and Stage-Level Design

TL;DR

PentestEval is a comprehensive benchmark that evaluates LLMs across all stages of penetration testing, revealing current limitations and emphasizing the need for modular, structured reasoning to improve automation reliability.

Contribution

Introduces PentestEval, the first detailed benchmark for stage-level evaluation of LLMs in penetration testing, highlighting performance gaps and guiding future improvements.

Findings

LLMs perform poorly across penetration testing stages.

End-to-end success rate of 31% for current LLM pipelines.

Autonomous agents fail almost entirely in penetration testing tasks.

Abstract

Penetration testing is essential for assessing and strengthening system security against real-world threats, yet traditional workflows remain highly manual, expertise-intensive, and difficult to scale. Although recent advances in Large Language Models (LLMs) offer promising opportunities for automation, existing applications rely on simplistic prompting without task decomposition or domain adaptation, resulting in unreliable black-box behavior and limited insight into model capabilities across penetration testing stages. To address this gap, we introduce PentestEval, the first comprehensive benchmark for evaluating LLMs across six decomposed penetration testing stages: Information Collection, Weakness Gathering and Filtering, Attack Decision-Making, Exploit Generation and Revision. PentestEval integrates expert-annotated ground truth with a fully automated evaluation pipeline across 346 tasks covering all stages in 12 realistic vulnerable scenarios. Our stage-level evaluation of 9 widely used LLMs reveals generally weak performance and distinct limitations across the stages of penetration-testing workflow. End-to-end pipelines reach only 31% success rate, and existing LLM-powered systems such as PentestGPT, PentestAgent, and VulnBot exhibit similar limitations, with autonomous agents failing almost entirely. These findings highlight that autonomous penetration testing demands stronger structured reasoning, where modularization enhances each individual stage and improves overall performance. PentestEval provides the foundational benchmark needed for future research on fine-grained, stage-level evaluation, paving the way toward more reliable LLM-based automation.