Every development team has a story like this. A pull request sits open for half a day. Not because the logic is broken or the API contract changed, but because two reviewers disagree on whether object literals need trailing commas. The thread grows. Someone posts a style guide link. Someone else counters with a different guide. By the time the code merges, everyone involved has lost context on the features they were actually building.
These fights are expensive. They burn hours of senior engineer time, create low-grade resentment between teammates, and train junior developers to believe that software engineering is mostly about winning arguments over semicolons. The worst part? The product does not care. Your users will never notice a space versus a tab. They will notice the bug you did not fix because you were busy debating quote marks.
Consistency does matter. A codebase that looks like it was written by one person is easier to read, easier to review, and easier to debug. The mistake is trying to enforce that consistency by hand.
Automate the Boring Stuff
The fix is straightforward. Remove human judgment from formatting entirely. Hand it off to tools that do not have egos and do not get tired.
Three tools handle this cleanly.
Prettier takes your code and reformats it automatically. It does not ask permission. You stop thinking about line lengths, quote styles, or how to break a long function signature across lines. You save the file. Prettier makes it consistent.
ESLint handles the problems Prettier will not touch. It catches unused variables, unreachable code, missing dependencies in React hooks, and patterns that historically lead to bugs. When configured correctly, it stays out of formatting and focuses on actual code quality.
Husky installs a pre-commit hook that blocks anything from entering your repository until the automated checks pass. It turns your Git pipeline into a gatekeeper instead of a suggestions box.
Together they form a tight loop. You write code however you want locally. When you commit, the tools clean and check it. Only then does the code leave your machine.
Why This Specific Stack Works
You can spend weeks tuning ESLint rules by hand. Resist that urge. The goal here is to stop debating style, not to create a new full-time job as a style guide curator.
Prettier is intentionally opinionated. It offers limited configuration options because every option is a future argument. The defaults are sane. Pick a small set of overrides, write them down once, and move on.
ESLint, left to its own devices, will try to enforce both code quality and formatting rules like semicolon usage and indent size. That creates friction with Prettier because both tools will try to edit the same characters. The package eslint-config-prettier solves this by disabling all ESLint rules that conflict with Prettier. This separation of duties is critical. Prettier owns the cosmetics. ESLint owns the logic.
Running checks only in continuous integration is too late. By the time the CI fails, you have already committed messy code, context-switched to another task, and possibly opened a pull request. Fixing it requires another commit, another push, another waiting cycle. Husky shortens that feedback loop to seconds. Lint-staged makes it fast by only running the tools against files you actually changed, rather than scanning the entire repository on every commit.
Setting It Up Step by Step
The following setup targets a modern JavaScript or React project, but the pattern translates to TypeScript, Vue, or Node with minor adjustments. Run each step from your project root.
Start by installing everything as dev dependencies:
npm install -D prettier eslint husky lint-staged eslint-config-prettier