Setting Up lint-staged with Pre-commit Hooks

Running linters on your entire codebase before every commit is slow and discouraging. lint-staged solves this by running your configured tools only on files that are staged in git, keeping the feedback loop tight and the developer experience smooth.

#// Installation

Install lint-staged alongside Husky for git hook management. In a monorepo using pnpm workspaces, install both at the root level so the hook applies to all packages.

The .lintstagedrc configuration file maps file glob patterns to the commands that should run against matching staged files. A typical setup runs ESLint with the --fix flag on TypeScript files and Prettier on everything else.

#// Configuration Strategy

Keep your lint-staged configuration focused on catching real issues. Run ESLint in quiet mode to suppress warnings and only block commits on errors. Run Prettier with the --write flag so formatting fixes are applied automatically rather than just reported.

For monorepos, use a root-level configuration that covers the common patterns, then add package-level overrides only when a specific package needs different rules. This prevents configuration sprawl and keeps the setup maintainable.

#// Handling Edge Cases

Some files should be excluded from pre-commit linting. Generated files, vendored dependencies, and build artifacts should be skipped using ignore patterns. Large files that take too long to lint can use a separate, lighter check.

If a pre-commit hook fails, the commit is blocked and the developer sees the errors. This is intentional, never add --no-verify as a workaround. Fix the underlying issue instead.