#6 The crash reporting only logs the last request sent

[nnp]

When a crash occurs, only the last fuzz request is logged. As a result, this is all the crash_reply.py script will resend. It might be nice to save the last sequence of requests or intelligently detect what sequence actually caused the crash.

This could be done by storing the last 10 or so requests and when a crash occurs, restarting the target and replaying them using a shrinking window e.g 1..10, 1..9 etc until the crash occurs again