How to customize build scripts in Replit

You customize build scripts in Replit by editing the Run and Build commands in the Shell → Replit Tools → Secrets & Config → .replit panel or directly modifying the .replit file and (if it exists) the replit.nix file. In most Repls, the .replit file controls exactly what command runs when you click the green Run button, and you’re free to replace that with your own build steps, scripts, and custom commands.

 

.replit: The main place to customize build scripts

 

The .replit file is a simple config file Replit reads to decide how your project is built and run. If you want custom build scripts, you usually edit the run or build entries inside it.

For example, if you want Replit to run a Node build step like npm run build before starting the server:

run = "npm run build && node server.js"  // build first, then start the server

Or if you prefer a dedicated build command:

build = "npm run build"  // only runs when you click 'Build' in the Replit UI
run = "node server.js"

Important: Not every template exposes the Build button, but run always works because it triggers when you hit the green Run button.

 

How Replit decides what to run

 

Replit automatically picks defaults based on your project type (Node, Python, etc.), but the moment you add a .replit file, you override that behavior. This gives you full control. Common cases:

• Node apps: default is node index.js, but you can swap it for webpack, vite, or any custom script.
• Python apps: default is python main.py, but you can hook in build steps, code generators, etc.
• React/Vite: Replit often defaults to npm start. If you need a production build, you can override it with your own command.

 

Example: Adding a true “build then run” flow for React/Vite

 

React/Vite Repls sometimes confuse beginners because npm run dev isn't a real "build". To create a proper build script:

build = "npm install && npm run build"  // ensure deps then build
run = "npm run preview"                 // run the built version

This gives you a more production-like environment inside Replit.

 

Using shell scripts if your build is complex

 

If your build steps grow messy, you can put them into a shell script like build.sh:

#!/bin/sh

npm install        // install deps
npm run generate   // maybe codegen step
npm run build      // your actual build

Then reference it in .replit:

build = "sh build.sh"
run = "node server.js"

This keeps your config file clean and easier for teammates to understand.

 

If your Repl uses Nix

 

In Nix-enabled Repls, replit.nix only controls system packages and environment setup, not the build script itself. So you almost always still customize build scripts inside .replit. The only time you edit replit.nix for build purposes is when you need extra tools (like ffmpeg, jq, gcc, etc.).

Example adding tools:

{ pkgs }: {
  deps = [
    pkgs.nodejs
    pkgs.ffmpeg
  ];
}

Then your .replit build script can rely on those tools.

 

Common pitfalls

 

• Forgetting install steps: Replit doesn’t automatically run npm install or pip install during builds unless you explicitly put them in your script.
• Using Windows commands: Replit uses Linux. So commands like dir or start won’t work.
• Long-running processes in build: Build commands should finish. Don’t put servers or watchers inside build.
• Confusing build and run: Build is optional; run is what executes when hitting the green Run button.

 

Simple template for most projects

 

If you're starting fresh and want something safe and predictable:

build = "npm install && npm run build"  // or pip install -r requirements.txt for Python
run = "npm start"                       // or python main.py

This pattern works well across Node, React, tooling-heavy apps, or mixed environments.

 

That’s everything you need to reliably customize build scripts in Replit without falling into the usual pitfalls.