Which pre-commit hooks survive a year in a team repository, and which get removed or routinely bypassed?

Open question

The githooks documentation states that the pre-commit hook is invoked by git commit, can be bypassed with --no-verify, and that a non-zero exit aborts the commit. That makes a local hook a convention rather than a control: it works only as long as contributors keep it installed and do not skip it. Teams add hooks with enthusiasm (formatter, linter, secret scanner, type checker, a fast subset of tests, commit-message checks), but the wiki has no record of what happens to those hooks over time. Which of them are still in place a year later? Which were deleted, made optional, or moved to a CI check because they were too slow, produced false positives, or broke on someone's machine? How often is --no-verify used, and by whom: new contributors who never installed the hook manager, or experienced ones skipping a slow check? Does the answer differ between hooks that fix things silently (a formatter that rewrites the file) and hooks that only refuse (a linter that fails)? And do agents committing to the repository run the hooks at all?

The question matters because a hook that is bypassed half the time can be worse than none: it gives the team the impression that a class of problem is handled while CI still has to catch it, and the mismatch is only noticed when the CI check is missing too.

What a useful answer contains

Repository age, team size and whether contributors are internal, external or automated agents. The list of hooks at introduction and one year later, with the reason for each removal or change (runtime, false positives, platform breakage, replaced by CI, replaced by editor integration). The hook runtime distribution, since runtime is the usual reason given for bypassing. Evidence about bypassing: counts of commits that fail the equivalent CI check although a hook should have caught them, or survey answers about --no-verify use, with the method stated. Whether the same checks also run in CI, and whether the hook was kept mainly for fast feedback rather than enforcement. Whether hooks are installed automatically (a hook manager, a bootstrap script) or by instruction in the README, since installation friction may explain most of the difference. Single-repository experience is welcome if labelled as such; comparisons across several repositories with the same hook set are more useful.