Clouseau: A Hierarchical Multi-Agent Approach for Autonomous Attack Investigation
Annual Computer Security Applications Conference (ACSAC) '25, Conference Paper (to appear)
Abdullah Aldaihan, Fahad Alotaibi, and Sergio Maffeis. Clouseau orchestrates a hierarchy of collaborating LLM agents that mirror
the way senior incident responders guide junior analysts. The system automatically triages alerts, correlates signals,
and reconstructs timelines, drastically reducing manual investigation effort.
Towards Reproducible Attack Scenarios With Automated Labeling
ACSAC '25 Poster Track
Jiayi Xie, Abdullah Aldaihan, and Sergio Maffeis. Demonstrates an automated labeling pipeline that uses large language models to
annotate complex attack traces, ensuring that adversarial datasets can be regenerated consistently for benchmarking and education.