The innovative technique introduced by ReActis based on a double characterization whose efficiency is proved by tests performed during project implementation. The project focused on a technique called fuzzing: fuzzers are algorithms that feed the tested software with random inputs and monitor if the application crashes as a result. Subsequently, it varies the input slightly and monitors if any new regions of the code are uncovered. In the process, the purpose of the fuzzer is to execute as much of the program code as possible and, ultimately, to test all parts of the program in order to check if a specific input causes them to crash.

Our technique shows that even in a binary-only scenario, where no source code is available for the system under test, our methods are powerful enough to compete with other approaches. This even holds for scenarios where hash checks are removed for the competing tools and they are provided with a proper input specification.