Sam now makes Husky 35,333 's pre-commit run lint-staged 14,736 , moves the commit-msg hook under Husky, drops the hand pre-commit hook (ESLint 39,810 catches syntax errors and more), commits, and tries the pipeline on a typo:
echo "npx lint-staged" > .husky/pre-commit
git mv .githooks/commit-msg .husky/commit-msg && git rm -q .githooks/pre-commit
git add . && git commit -qm "Lint staged JavaScript with ESLint before each commit"
git push -q
# Edit app.js: a typo, "res.json(row)" in GET /api/genres
git commit -qam "Return the genre list"; echo "exit status $?"
git restore --staged --worktree app.jsOutput
... ⋯ Reverting to original state because of errors… ... ✖ eslint --fix --max-warnings=0: /home/dev/v5-ch2/booknest/app.js 69:15 error 'rows' is assigned a value but never used no-unused-vars 72:16 error 'row' is not defined no-undef ✖ 2 problems (2 errors, 0 warnings) husky - pre-commit script failed (code 1) exit status 1
The first commit passed; lint-staged linted its one staged JavaScript file, eslint.config.js. The second was stopped by two real errors, the kind of typo that otherwise surfaces as a 500 error in production. A fresh clone now gets both hooks from npm 2,036 install alone, and the configured whitespace hook still runs from Sam's local configuration. Keep this stage under a few seconds; slow checks such as BookNest's database tests belong in pre-push or, better, in CI (GitHub), where nobody can skip them.