Environment Variables

Create T3 App uses its own package @t3-oss/env-nextjs↗ along with zod↗ under the hood for validating environment variables at runtime and buildtime by providing a simple logic in src/env.mjs.

env.mjs

TLDR; If you want to add a new environment variable, you must add a validator for it in src/env.mjs, and then add the KV-pair in .env

import { createEnv } from "@t3-oss/env-nextjs";
import { z } from "zod";

export const env = createEnv({
  server: {
    NODE_ENV: z.enum(["development", "test", "production"]),
  },
  client: {
    // NEXT_PUBLIC_CLIENTVAR: z.string(),
  },
  runtimeEnv: {
    NODE_ENV: process.env.NODE_ENV,
  },
});

T3 Env brings a new concept across the createEnv, which is responsible for creating the schema and will include the main validation logic for client and server-sided environment variables.

Using Environment Variables

When you want to use your environment variables, you can import them from the created env.mjs and use them as you would normally do. If you import this on the client and try accessing a server-side environment variable, you will get a runtime error.

import { env } from "../../env.mjs";

// `env` is fully typesafe and provides autocompletion
const dbUrl = env.DATABASE_URL;

import { env } from "../env.mjs";

// ❌ This will throw a runtime error
const dbUrl = env.DATABASE_URL;

// ✅ This is fine
const wsKey = env.NEXT_PUBLIC_WS_KEY;

.env.example

Since the default .env file is not committed to version control, we have also included a .env.example file, in which you can optionally keep a copy of your .env file with any secrets removed. This is not required, but we recommend keeping the example up to date to make it as easy as possible for contributors to get started with their environment.

Some frameworks and build tools, like Next.js, suggest that you store secrets in a .env.local file and commit .env files to your project. This is not recommended, as it could make it easy to accidentally commit secrets to your project. Instead, we recommend that you store secrets in .env, keep your .env file in your .gitignore and only commit .env.example files to your project.

Adding Environment Variables

To ensure your build never completes without the environment variables the project needs, you will need to add new environment variables in two locations:

📄 .env: Enter your environment variable like you would normally do in a .env file, i.e. KEY=VALUE

📄 env.mjs: Add the appropriate validation logic for the environment variables by defining a Zod schema inside createEnv for each one, e.g. KEY: z.string(). Besides that, make sure to destruct them in the runtimeEnv option, e.g.: KEY: process.env.KEY

Optionally, you can also keep .env.example updated:

📄 .env.example: Enter your environment variable, but be sure to not include the value if it is secret, i.e. KEY=VALUE or KEY=

Example

I want to add my Twitter API Token as a server-side environment variable

1. Add the environment variable to .env:

TWITTER_API_TOKEN=1234567890

2. Add the environment variable to env.mjs:

import { createEnv } from "@t3-oss/env-nextjs";
import { z } from "zod";

export const env = createEnv({
  server: {
    TWITTER_API_TOKEN: z.string(),
  },
  // ...
  runtimeEnv: {
    // ...
    TWITTER_API_TOKEN: process.env.TWITTER_API_TOKEN,
  },
});

3. Optional: Add the environment variable to .env.example and make sure not to include the secret in the runtimeEnv option

TWITTER_API_TOKEN=