Skip to main content
Courser’s backend reads configuration from environment variables at startup. In development, set them in a .env file in the repository root. In production, inject them through your host’s environment or secret manager.
When NODE_ENV is set to production, the backend skips loading .env via dotenv. All variables must be present in the environment before the process starts.

​
Backend variables

Create a .env file in the repository root (the same directory as app.js).

​
Database

​
MONGO_URI
string
required
MongoDB connection string. Used by Mongoose to connect to your database on startup.For MongoDB Atlas, copy the connection string from Database → Connect → Drivers. Replace <password> with your actual password and optionally append a database name.
MONGO_URI=mongodb+srv://username:[email protected]/courser

​
AI services

​
OPENAI_API_KEY
string
required
Platform-default OpenAI API key. Used for:
  • Embeddings — text-embedding-ada-002 model, called when a professor adds lecture videos
  • Chat completions — gpt-3.5-turbo-16k model, called on every student question
This key is used for courses that do not have a professor-supplied API key. Obtain one from the OpenAI platform.
OPENAI_API_KEY=sk-...
​
PINECONE_API_KEY
string
required
Pinecone API key. Used to read and write lecture embeddings in the vector index.The index name is hardcoded as courser in classes/CourserAIAssistant.js. You must create an index with that name in your Pinecone project before starting the backend. See Self-hosting overview for index creation settings.Obtain your key from the Pinecone console under API Keys.
PINECONE_API_KEY=pcsk_...

​
Authentication

​
JWT_PRIVATE_KEY
string
required
Secret key used to sign and verify JWT tokens for professor sessions. Set this to a long, random string — at least 32 characters.
Use a strong, unpredictable value in production. Anyone who obtains this key can forge valid session tokens for any professor account.
Generate a suitable value with:
node -e "console.log(require('crypto').randomBytes(48).toString('hex'))"

JWT_PRIVATE_KEY=a3f9c2d8e1b74a6f...

​
File storage

​
CLOUDINARY_URL
string
required
Cloudinary connection URL, used for uploading chatbot background images. This is the all-in-one credential URL that Cloudinary provides in your dashboard.Find it in the Cloudinary console under Dashboard → API Keys.
CLOUDINARY_URL=cloudinary://API_KEY:API_SECRET@CLOUD_NAME
Alternatively, you can configure Cloudinary using individual variables (CLOUDINARY_CLOUD_NAME, CLOUDINARY_API_KEY, CLOUDINARY_API_SECRET) if your Cloudinary SDK version supports it, but CLOUDINARY_URL is the format used in the current source.

​
Server

​
PORT
number
default:"8000"
Port the Express server listens on. Optional — defaults to 8000 if not set.
PORT=8000
​
NODE_ENV
string
default:"development"
Runtime environment. Set to production in production deployments.When set to production, the backend skips calling require('dotenv').config(), so it will not read a .env file. All variables must already be in the environment.
NODE_ENV=production

​
Firebase configuration

Firebase Auth is configured differently from the variables above. The Firebase client config is currently hardcoded as an object in endpoints/auth.js: 
// endpoints/auth.js
const firebaseConfig = {
    apiKey: "...",
    authDomain: "...",
    projectId: "...",
    storageBucket: "...",
    messagingSenderId: "...",
    appId: "...",
};
For a self-hosted deployment, replace this object with the configuration from your own Firebase project. You can find it in the Firebase console under Project Settings → Your apps → Web app → SDK setup and configuration.
Firebase client config values (API key, project ID, etc.) are designed to be public — they identify your Firebase project but do not grant privileged access on their own. Security is enforced by Firebase Authentication rules and your backend’s JWT verification. That said, use your own project’s config in any self-hosted deployment so that accounts belong to your project.

​
Frontend variables

Create a .env.local file in the client directory for local development. In production, set these through your hosting provider.
​
NEXT_PUBLIC_API_URL
string
required
Full base URL of your backend API, without a trailing slash. The Next.js frontend uses this to construct all API requests.
# Local development
NEXT_PUBLIC_API_URL=http://localhost:8000

# Production
NEXT_PUBLIC_API_URL=https://api.your-domain.com
Because this variable is prefixed with NEXT_PUBLIC_, Next.js inlines it into the browser bundle at build time. Changing it requires a rebuild.

​
Sample .env file

Copy this template to the repository root as .env and fill in your values:
.env
# -----------------------------------------------
# Courser backend — environment variables
# -----------------------------------------------

# MongoDB connection string
MONGO_URI=mongodb+srv://username:[email protected]/courser

# OpenAI API key (embeddings + chat completions)
OPENAI_API_KEY=sk-...

# Pinecone API key
# Index name must be "courser" — create it in the Pinecone console
PINECONE_API_KEY=pcsk_...

# JWT secret — use a long random string, keep this private
JWT_PRIVATE_KEY=replace-with-a-long-random-secret

# Cloudinary — background image uploads
CLOUDINARY_URL=cloudinary://api_key:api_secret@cloud_name

# Server port (optional, defaults to 8000)
# PORT=8000

# Set to "production" in production deployments
# NODE_ENV=production
And in client/.env.local:
client/.env.local
# Backend API base URL (no trailing slash)
NEXT_PUBLIC_API_URL=http://localhost:8000

​
Variable summary

VariableRequiredDefaultUsed in
MONGO_URIYes—app.js
OPENAI_API_KEYYes—classes/CourserAIAssistant.js
PINECONE_API_KEYYes—classes/CourserAIAssistant.js
JWT_PRIVATE_KEYYes—middleware.js
CLOUDINARY_URLYes—endpoints/course.js
PORTNo8000app.js
NODE_ENVNodevelopmentapp.js
NEXT_PUBLIC_API_URLYes—Next.js frontend

Build docs developers (and LLMs) love

Get started for freeTalk to us