CyberForge: Verified Vulnerability Injection at Repository Level for Cybersecurity Agent Training

Despite recent advances, frontier large language model (LLM) agents remain limited in discovering and patching complex vulnerabilities in real-world software. Generally available agents can already aid attackers, who only need to find one exploitable weakness, while defenders must continuously identify and patch all vulnerabilities across fast-growing codebases. Stronger defensive agents would help close this gap, yet the scarcity of security training data with reproducible build and execution environments remains a bottleneck. We present CyberForge, a framework that synthesizes executable, repository-level security training data by injecting vulnerabilities into real C/C++ projects. It validates each instance dynamically: the injected build must pass the project's unit tests, and generated proof-of-vulnerability (PoV) must trigger on the injected build and not on the clean one. CyberForge is not limited by the availability of disclosed vulnerabilities, therefore it can scale in comparison to data augmentation techniques which rely on historic CVE data. The resulting corpus holds 1034 validated vulnerabilities across 80 projects and 63 weakness categories, with edit locality similar to real CVE patches under a real-versus-real noise floor. Fine-tuning on trajectories collected over this corpus improves SEC-bench patch repair by +3.3 to +14.7 points, in all six configurations of three model scales and two teachers, with the 31B student reaching its GPT-5.4-mini teacher, 72.7% against 74.0%. These gains generalize out of distribution to PatchEval, a corpus containing other programming languages, where every configuration also improves and the 31B student passes its teacher.

Toward Scalable Automated Repository-Level Datasets for Software Vulnerability Detection

Software vulnerabilities continue to grow in volume and remain difficult to detect in practice. Although learning-based vulnerability detection has progressed, existing benchmarks are largely function-centric and fail to capture realistic, executable, interprocedural settings. Recent repo-level security benchmarks demonstrate the importance of realistic environments, but their manual curation limits scale. This doctoral research proposes an automated benchmark generator that injects realistic vulnerabilities into real-world repositories and synthesizes reproducible proof-of-vulnerability (PoV) exploits, enabling precisely labeled datasets for training and evaluating repo-level vulnerability detection agents. We further investigate an adversarial co-evolution loop between injection and detection agents to improve robustness under realistic constraints.

AI Agentic Vulnerability Injection And Transformation with Optimized Reasoning

The increasing complexity of software systems and the sophistication of cyber-attacks have underscored the critical need for effective automated vulnerability detection and repair systems. Traditional methods, such as static program analysis, face significant challenges related to scalability, adaptability, and high false-positive and false-negative rates. AI-driven approaches, particularly those using machine learning and deep learning models, show promise but are heavily reliant on the quality and quantity of training data. This paper introduces a novel framework designed to automatically introduce realistic, category-specific vulnerabilities into secure C/C++ codebases to generate datasets. The proposed approach coordinates multiple AI agents that simulate expert reasoning, along with function agents and traditional code analysis tools. It leverages Retrieval-Augmented Generation for contextual grounding and employs Low-Rank approximation of weights for efficient model fine-tuning. Our experimental study on 116 code samples from three different benchmarks suggests that our approach outperforms other techniques with regard to dataset accuracy, achieving between 89\% and 95\% success rates in injecting vulnerabilities at function level.