Quick Start

# Set Node version
nvm use

# Install yarn v1
npm i -g yarn

# Install dependencies
yarn

# Setup environment secrets
# Create api/.env.local with your OAuth credentials
cp api/.env api/.env.local
# Edit api/.env.local and add your GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET,
# and generate new ACCESS_TOKEN_SECRET and REFRESH_TOKEN_SECRET with:
# node -e "console.log(require('crypto').randomBytes(64).toString('hex'))"

# Start docker, create & seed DBs
docker compose up -d

# Run database migrations
yarn migrate:oauth

# Build common types
yarn build

# Start API (terminal 1)
yarn run start:api

# Start web app (terminal 2)
yarn run start:web

Environment Setup

Required: api/.env.local

Create api/.env.local (gitignored) with your actual secrets:

# Copy template
cp api/.env api/.env.local

Then edit api/.env.local and set:

1. Google OAuth credentials (from Google Cloud Console)

   • GOOGLE_CLIENT_ID
   • GOOGLE_CLIENT_SECRET

2. JWT secrets (generate new random secrets):

   # Generate ACCESS_TOKEN_SECRET
   node -e "console.log(require('crypto').randomBytes(64).toString('hex'))"
   
   # Generate REFRESH_TOKEN_SECRET
   node -e "console.log(require('crypto').randomBytes(64).toString('hex'))"

3. OpenAI API Key (for journal analysis feature)

   • OPENAI_API_KEY - Your OpenAI API key from https://platform.openai.com/api-keys

The .env file contains default/placeholder values for version control. The .env.local file (gitignored) overrides with actual secrets.

Logs

# follow logs for postgres
docker-compose logs -f postgres

Testing

API Unit Tests

yarn test:api

API E2E Tests

Requires API running on port 3000:

# Terminal 1: API server
yarn start:api

# Terminal 2: Run tests
yarn test:api:e2e

Frontend Cypress Tests

Requires both API (port 3000) and web app (port 4200) running:

# Terminal 1: Web dev server
yarn start:web

# Terminal 2: Run Cypress
yarn test:web

Prompts Service (Daily Reflection Questions)

Overview

The prompts service is a daily cronjob that generates personalized reflection questions for users based on their journal analysis and recent entries. When users visit the "Create New Journal" page, they're presented with a thoughtful prompt in a modal to inspire their writing.

How It Works

1. Daily Generation: A Kubernetes CronJob runs every morning at 2:00 AM
2. User Scanning: Iterates through all users in the system
3. Prompt Creation: For each user with analysis data:
   • Fetches their latest user analysis (themes, patterns)
   • Retrieves their most recent journal entry
   • Calls OpenAI GPT-4o-mini to generate a single reflective question
   • Stores the prompt in the user_prompts database table
4. Modal Presentation: When a user visits /journals/new, the modal displays their latest unacknowledged prompt

Database Schema

The user_prompts table stores generated prompts:

CREATE TABLE user_prompts (
  "promptId" VARCHAR(255) PRIMARY KEY,
  "userId" VARCHAR(255) NOT NULL,
  "prompt" TEXT NOT NULL,
  "acknowledgedAt" TIMESTAMP,
  "createdAt" TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
  FOREIGN KEY ("userId") REFERENCES "user"("userId") ON DELETE CASCADE
);

Modal Behavior

• Dismiss (click outside): Modal closes but will reappear on next visit (stored in localStorage)
• Acknowledge (click button): Modal closes permanently, calls API to mark prompt as acknowledged

Manual Testing

Run the prompts generation script manually:

# Requires API and database to be running
yarn start:prompts

This will:

• Connect to the database
• Scan all users
• Generate prompts for users with analysis data
• Log success/skip/error counts
• Exit when complete

Kubernetes Deployment

The CronJob is configured in k8s/base/prompts/cronjob.yaml:

• Schedule: 0 2 * * * (2:00 AM daily)
• Concurrency Policy: Forbid (prevents overlapping runs)
• Job History: Keeps last 3 successful and 1 failed job
• Timeout: 30 minutes
• Resources: 128Mi-512Mi memory, 100m-500m CPU

Build and deploy:

# Build Docker image
docker build -t journal-prompts:local -f prompts/Dockerfile .

# Deploy to Kubernetes (if using minikube)
kubectl apply -f k8s/base/prompts/cronjob.yaml

# Manually trigger a job for testing
kubectl create job --from=cronjob/prompts-generator manual-prompts-test -n journal

# Check job status
kubectl get jobs -n journal
kubectl logs job/manual-prompts-test -n journal