LibAFLstar
LibAFLstar is a fast and state-aware protocol fuzzer that addresses the challenges of fuzzing stateful software systems. Unlike stateless approaches, LibAFLstar efficiently explores state models, focuses on interesting states, tracks relevant messages per state, and handles expensive system restarts.
Key Features
- Partial Message Sequences: Efficiently explores protocol state spaces
- Novel State Scheduler: Intelligently prioritizes states for testing
- State-Aware Queues and Bitmaps: Optimized data structures for stateful fuzzing
- Persistent Mode: Reduces overhead from system restarts
Performance
Our evaluation shows that LibAFLstar is more than 30× faster than state-of-the-art competitors (AFLNet and ChatAFL) and achieves 1.4× more coverage on average when testing protocol implementations (FTP, RTSP, and HTTP).
Publication
This work has been accepted at ESORICS 2025 and is co-authored by Cristian Daniele, Timme Bethe, Marcello Maugeri, Andrea Continella, and Erik Poll.
Next Steps
Current development focuses on improving the reproducibility of experimental results and expanding support for additional protocol types.
Get Involved
Interested in contributing or collaborating? Feel free to contact me or check out the GitHub repository for more details.
