Do you practice mutation testing and does it actually improve your test suite?

Mutation testing (mutmut for Python, Stryker for JS) sounds powerful in theory: introduce small code changes and verify tests catch them. I tried `mutmut` on one of our services. It found 3 tests that were passing without catching real bugs (always-true assertions). That's valuable. But it also: - Took 40 minutes to run on a 3,000-line module - Generated 600+ mutants, 90% of which were caught - Produced a lot of noise around logging and formatting code Is mutation testing practical for full suites, or is it better used surgically on critical modules? Has it caught anything that mattered in a real codebase for you?