Guided Reasoning in LLM-Driven Penetration Testing Using Structured Attack Trees

2025-09-09

AI Summary

Large language models (LLMs) exhibit hallucination, invalid actions, and cyclic responses in autonomous penetration testing due to unstructured reasoning. Method: This paper proposes a deterministic task-tree-guided reasoning framework grounded in the MITRE ATT&CK matrix, explicitly embedding structured adversarial knowledge into LLM decision-making to constrain generation along executable, real-world attack chains. Contribution/Results: The framework significantly suppresses hallucination while improving path validity and interpretability. Evaluated on Llama-3-8B, Gemini-1.5, and GPT-4 across 103 realistic subtasks, it achieves up to 78.6% task completion—nearly 5× more efficient than self-directed baselines—and reduces API call consumption by two-thirds. Its core innovation lies in pioneering the use of ATT&CK-driven attack trees as a controllable reasoning skeleton for knowledge-guided security agents.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents
• Natural Language Processing: (Large) Language Models

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent advances in Large Language Models (LLMs) have driven interest in automating cybersecurity penetration testing workflows, offering the promise of faster and more consistent vulnerability assessment for enterprise systems. Existing LLM agents for penetration testing primarily rely on self-guided reasoning, which can produce inaccurate or hallucinated procedural steps. As a result, the LLM agent may undertake unproductive actions, such as exploiting unused software libraries or generating cyclical responses that repeat prior tactics. In this work, we propose a guided reasoning pipeline for penetration testing LLM agents that incorporates a deterministic task tree built from the MITRE ATT&CK Matrix, a proven penetration testing kll chain, to constrain the LLM's reaoning process to explicitly defined tactics, techniques, and procedures. This anchors reasoning in proven penetration testing methodologies and filters out ineffective actions by guiding the agent towards more productive attack procedures. To evaluate our approach, we built an automated penetration testing LLM agent using three LLMs (Llama-3-8B, Gemini-1.5, and GPT-4) and applied it to navigate 10 HackTheBox cybersecurity exercises with 103 discrete subtasks representing real-world cyberattack scenarios. Our proposed reasoning pipeline guided the LLM agent through 71.8%, 72.8%, and 78.6% of subtasks using Llama-3-8B, Gemini-1.5, and GPT-4, respectively. Comparatively, the state-of-the-art LLM penetration testing tool using self-guided reasoning completed only 13.5%, 16.5%, and 75.7% of subtasks and required 86.2%, 118.7%, and 205.9% more model queries. This suggests that incorporating a deterministic task tree into LLM reasoning pipelines can enhance the accuracy and efficiency of automated cybersecurity assessments

Problem

• Automating cybersecurity penetration testing using LLMs
• Reducing inaccurate or hallucinated procedural steps in LLM agents
• Improving efficiency and accuracy of automated vulnerability assessments

Innovation

• Guided reasoning pipeline with structured attack trees
• Deterministic task tree from MITRE ATT&CK Matrix
• Constrains LLM to proven tactics and procedures