This project explores how contextual knowledge can be systematically modeled and embedded into AI agents operating in cybersecurity environments. The project focuses on understanding how representations such as ontologies, knowledge graphs, and structured threat intelligence can shape agent reasoning, improve accuracy, and reduce hallucinations. Students will investigate methods to identify adversary behaviors, co-occurring attack techniques, and environmental signals, while examining how these can be integrated into agent workflows for tasks such as threat emulation, detection engineering, and threat intelligence sharing.
The project also considers challenges of explainability, adaptability, and reliability in autonomous security systems.
Further Reading:
- Context Engineering Handbook: https://github.com/davidkimai/Context-Engineering
- Prompting guide overview: https://www.promptingguide.ai/guides/context-engineering-guide