Crash triage is the process of sorting, reproducing, and prioritizing program failures found during fuzzing. A fuzzer may trigger hundreds or thousands of crashes, but many are duplicates, expected aborts, or symptoms of the same root bug. Triage turns that noisy output into actionable security work by grouping crashes, checking whether they are real defects, and determining which ones are most urgent.
This matters because fuzzing only reduces risk if the team can quickly turn crashes into fixes. In practice, triage often means inspecting stack traces, minimizing test cases, reproducing the failure reliably, and deciding whether it is a security issue such as memory corruption, denial of service, or a parser bug. Good crash triage is a defense workflow: it helps engineers focus on exploitable or high-impact failures first, avoid wasting time on duplicates, and keep continuous fuzzing useful as code changes over time.