behave

behave records how a command-line tool behaves (flags, stdin, files, exit codes, stdout/stderr, edge cases) by generating test cases from its --help and man page, stores them as readable YAML, and replays them against another implementation to show, per flag, where a rewrite differs. It is aimed at memory-safe rewrites of core tools (e.g. uutils coreutils, gojq) and at regression testing of CLIs. Status: early (v0.1.x). Results so far include compatibility bugs reported to uutils/coreutils with fixes merged upstream (#15345, #15381).