cs.CRJul 29, 2025

Cyber-Zero: Training Cybersecurity Agents without Runtime

Authors: Terry Yue Zhuo, Dingmin Wang, Hantian Ding, Varun Kumar, Zijian Wang

Organizations: Monash University · AWS AI Labs

Abstract

Large Language Models (LLMs) have achieved remarkable success in software engineering tasks when trained with executable runtime environments, particularly in resolving GitHub issues. However, such runtime environments are often unavailable in other domains, especially cybersecurity, where challenge configurations and execution contexts are ephemeral or restricted. We present Cyber-Zero, the first runtime-free framework for synthesizing high-quality agent trajectories to train cybersecurity LLMs. Cyber-Zero leverages publicly available CTF writeups and employs persona-driven LLM simulation to reverse-engineer runtime behaviors and generate realistic, long-horizon interaction sequences without actual environments. Using trajectories synthesized by Cyber-Zero, we train LLM-based agents that achieve up to 11.4% absolute performance gains over baseline models on three prominent CTF benchmarks: InterCode-CTF, NYU CTF Bench, and Cybench. Our best model, Cyber-Zero-32B, establishes new state-of-the-art performance among open-weight models, matching the capabilities of proprietary systems like DeepSeek-V3-0324 and Claude-3.5-Sonnet while offering superior cost-effectiveness, and demonstrating that runtime-free trajectory synthesis can effectively democratize the development of state-of-the-art cybersecurity agents.

Figures & tables

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Aug 24, 2026

CyberFactory: Scaling Cyber Security Capabilities with Instances from the Wild

As large language models (LLMs) continue to advance in coding capabilities, their potential in cybersecurity has drawn increasing research attention, with closed-source LLMs (e.g., Mythos) delivering advanced cybersecurity capabilities. However, existing open-source efforts remain limited: frontier open-weight models do not provide reproducible cybersecurity training solutions, open-source training solutions focus on isolated tasks and lack scalable agentic data, and scaling agentic rollouts requires strong domain priors. In this work, we introduce CyberFactory, a unified open-source framework that connects data construction, trajectory synthesis, and model training across proof-of-concept (PoC) generation, vulnerability patching, and cybersecurity question answering (CyberQA). CyberFactory transforms public vulnerability artifacts, including CVEs from the wild, into executable and verifiable task instances. It further uses a reusable vulnerability-analysis skill to guide the teacher through source inspection, problem solving with domain prior, and evidence-based validation. The resulting supervision is agentic: the model interacts with tools and target environments and revises its solutions according to execution feedback. Using these trajectories, we train and release \modelname\footnote{Aegis is, in Greek mythology, the protective shield of Zeus and Athena; the name reflects the model's defensive, security-oriented purpose.}, which internalizes the skill-guided procedure without requiring the skill at inference time. On CyberGym, \modelname reaches 52.4% Pass@1 under a one-hour budget, improving over its Qwen~3.5 base model by +22.8 points and outperforming the evaluated general-purpose backbones under the same scaffold.
Mar 2, 2026

ZeroDayBench: Evaluating LLM Agents on Unseen Zero-Day Vulnerabilities for Cyberdefense

Large language models (LLMs) are increasingly being deployed as software engineering agents that autonomously contribute to repositories. A major benefit these agents present is their ability to find and patch security vulnerabilities in the codebases they oversee. To estimate the capability of agents in this domain, we introduce ZeroDayBench, a benchmark where LLM agents find and patch 22 novel critical vulnerabilities in open-source codebases. We focus our efforts on three popular frontier agentic LLMs: GPT-5.2, Claude Sonnet 4.5, and Grok 4.1. We find that frontier LLMs are not yet capable of autonomously solving our tasks and observe some behavioral patterns that suggest how these models can be improved in the domain of proactive cyberdefense.
Mar 23, 2026

STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving

Autonomous cyber operations increasingly depend on agents that can chain reconnaissance, exploitation, and analysis across many tools while maintaining state over long task horizons. Large Language Models (LLMs) are promising orchestrators for this workflow, but naive agents that pipe model output directly into a shell are brittle: they emit malformed commands, hallucinate non-existent command-line options or packages, and lose track of earlier findings as the context window fills. We present STRIATUM-CTF, a modular agentic framework for Capture-the-Flag (CTF) solving that mediates between LLM reasoning and tool execution using the Model Context Protocol (MCP). Offensive-security tools are exposed as strictly typed, schema-validated functions; the protocol layer rejects out-of-schema tool calls before execution and returns structured observations that the agent uses to self-correct. We evaluate the system in two ways. In a controlled study of 180 runs across 15 CTF challenges under four levels of attached procedural documentation, the system succeeds on 77.8-86.7% of attempts; notably, a schema-only configuration with no procedural documentation already reaches 77.8%, and no condition differs significantly from another. In a single live, multi-team CTF at Florida State University, the system placed first against a field of more than 30 registered teams -- a single, non-replicated event decided by a 10-point margin over the runner-up, and therefore a demonstration of feasibility rather than of superiority over human experts. Because all four ablation conditions share the same tool layer, our controlled study isolates documentation volume rather than the protocol layer itself; we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect.