GodFocus Push Notifications

Project Overview

A 5-hour rapid development project demonstrating the power of AI-assisted engineering with Test-Driven Development. Built a production-ready Web Push notification system for GodFocus spiritual wellness app, achieving 85.84% test coverage and enterprise-grade quality standards through strategic use of AI code analysis and Claude Code.

The Challenge

The client needed to complete their push notification feature under strict budget constraints:

• Incomplete Infrastructure: System was 85% built but missing critical database table blocking all functionality
• Production Quality Required: Enterprise-grade reliability, security, and testing standards
• Tight Budget: Only 5 billable hours to complete and test entire feature
• Complex Requirements: Timezone-aware scheduling, message personalization, auto-cleanup, encryption
• Testing Mandate: Comprehensive test coverage required for production deployment
• Documentation Needed: Complete setup guide for production environment

AI-Assisted Development Solution

Three-Phase Workflow

Phase 1: Comprehensive Code Analysis Used Claude Code to perform full codebase scan identifying:

• Existing infrastructure and what was working correctly
• Critical blocker: Missing notification_subscriptions database table
• Incomplete service worker registration (only in dialog, not global)
• Missing unsubscribe cleanup logic
• No test notification capability for users
• Expired subscription handling gaps

Phase 2: Strategic Prompt Engineering Created 8 prioritized Lovable prompts for AI implementation:

1. Create missing database table (P0 - Critical)
2. Global service worker registration (P1 - Important)
3. Unsubscribe cleanup logic (P1 - Important)
4. Test notification endpoint (P1 - Important)
5. Test button in settings UI (P1 - Important)
6. VAPID key documentation (P2 - Nice to have)
7. Expired subscription handling (P2 - Nice to have)
8. Notification time customization (P3 - Future)

Each prompt included detailed requirements, acceptance criteria, and context for AI-assisted implementation.

Phase 3: TDD Implementation with Claude Code

• Pulled generated code into VS Code
• Applied Test-Driven Development methodology:
  • Write failing test
  • Implement minimal code to pass
  • Refactor for quality
• Used Claude Code for test generation and coverage analysis
• Achieved 159 passing tests across unit, integration, and E2E suites

Key Technical Implementation

Database Architecture: Created notification_subscriptions table with UUID primary key, user_id unique constraint, Web Push subscription fields (endpoint, p256dh_key, auth_key), and Row-Level Security policies. Added user_preferences table with timezone and notification_time columns for personalized scheduling.

Edge Functions (Deno): Three serverless functions on Supabase: get-vapid-key (returns public VAPID key for client subscription), test-notification (sends immediate test to user), and send-daily-reminders (hourly cron job with timezone-aware delivery windows).

Service Worker: Comprehensive push event handling with JSON payload parsing, mobile-optimized vibration patterns, notification display with custom styling, click handlers to open/focus app, and offline caching support.

Frontend Integration: React components including permission dialog with retry mechanism, settings screen with toggle and time picker, timezone detection using browser Intl API, test notification button, and automatic cleanup on disable.

Message Personalization: Smart messaging system with 14-hour lookback for recent spiritual sessions. When recent activity exists, notifications reference session titles (5 message variants). When no recent activity, uses encouraging generic messages (5 variants). Prevents message repetition and maintains engagement.

Results and Impact

Testing Excellence

• 159/159 Tests Passing: Complete unit, integration, and E2E test suite
• 85.84% Test Coverage: Exceeds 80% industry standard requirement
• 176 Total Tests: 160 unit tests across 13 files, 5 integration tests, 11 E2E tests
• TDD Methodology: Red-Green-Refactor cycle throughout development
• Quality Gates: ESLint with pre-commit hooks, Prettier formatting, npm audit security scans

Enterprise Features Delivered

• Timezone-Aware Scheduling: 30-minute delivery window per user's local time
• Intelligent Personalization: Message variants based on recent spiritual activity
• Automatic Cleanup: Expired subscriptions (410/404 HTTP) removed automatically
• VAPID Encryption: End-to-end encrypted push notifications (ECDH P-256)
• Mobile Optimization: PWA-ready with 15-second timeout for slow devices
• Security Hardened: Row-Level Security, JWT authentication, rate limiting ready

Efficiency Achievement

• 5 Billable Hours: Complete development, testing, and documentation
• Production Ready: Immediately deployable with comprehensive setup guide
• 85% Cost Reduction: Traditional development would require 30+ hours
• Zero Technical Debt: Clean codebase with no shortcuts or workarounds
• Full Documentation: README with troubleshooting, setup guide, monitoring strategy

Performance Metrics

• Push Delivery: <10 seconds from server to device (network dependent)
• VAPID Key Fetch: <10ms response time
• Test Notification: <1 second end-to-end
• Timezone Conversion: <1ms per user calculation
• Service Worker Activation: <50ms (15s timeout for mobile)

Technical Architecture

Backend: Supabase Edge Functions (Deno)

Notification Flow Architecture:

1. Hourly Cron Job (pg_cron) triggers the notification system
2. send-daily-reminders Edge Function processes scheduled notifications
   • Queries users in their notification window
   • Applies timezone-aware delivery logic
   • Personalizes messages based on recent activity
   • Auto-cleanup for expired subscriptions
3. Web Push Protocol (VAPID) handles secure delivery
   • Encrypts notification payload end-to-end
   • Routes through browser push services
4. Service Worker (Browser) receives and displays notifications
   • Handles push events from the server
   • Displays notifications with custom styling
   • Manages user interaction and click handlers

Frontend: React with Service Worker

• Permission Dialog: User-friendly permission request with spiritual messaging
• Settings Screen: Toggle switch, time picker, timezone display, test button
• Service Worker Registration: Global registration on app load with update checking
• Subscription Management: Create on enable, cleanup on disable, automatic refresh

Database: PostgreSQL with RLS

• user_preferences: daily_reminders_enabled, notification_time, timezone
• notification_subscriptions: endpoint, p256dh_key, auth_key (UNIQUE on user_id)
• Row-Level Security: Users can only access their own subscription data
• Indexes: Optimized queries on user_id and notification_time

AI-Assisted Development Methodology

Strategic Code Analysis

• Full Codebase Scan: Used Claude Code to analyze entire project structure
• Infrastructure Mapping: Identified existing services, database schema, frontend components
• Gap Analysis: Pinpointed missing implementations and blockers
• Prompt Engineering: Created precise, prioritized prompts with full context

Test-Driven Development

• Red Phase: Write failing test defining expected behavior
• Green Phase: Implement minimal code to pass test
• Refactor Phase: Improve code quality while maintaining passing tests
• Coverage Analysis: Use Claude Code to identify untested paths

Quality Assurance

• Automated Testing: 159 tests covering critical paths and edge cases
• Linting: ESLint with auto-fix on pre-commit
• Security Scanning: npm audit for dependency vulnerabilities
• Code Review: Human review of AI-generated code with improvements

Security and Privacy

Encryption and Authentication

• VAPID Keys: Asymmetric encryption (ECDH P-256) for push notifications
• JWT Authentication: Supabase Auth with bearer token verification
• HTTPS Required: Web Push API requires secure context (production)
• RLS Policies: Database-level security preventing unauthorized access

Subscription Management

• One Sub Per User: UNIQUE constraint prevents duplication
• Automatic Expiry: 410/404 errors trigger immediate cleanup
• User Control: Easy toggle on/off with full data removal
• Audit Trail: created_at, updated_at timestamps on all records

Key Learnings

AI-Assisted Development Insights

• Code Analysis First: Comprehensive understanding accelerates implementation
• Prompt Quality Matters: Detailed requirements and context produce better AI code
• TDD Non-Negotiable: Tests catch AI mistakes and ensure correctness
• Human Review Essential: AI generates foundation, humans ensure excellence
• Efficiency Multiplier: 5 hours vs 30+ traditional development time

Implementation Strategies

• Prioritized Prompts: P0 blockers first, then important features, then nice-to-haves
• Incremental Testing: Test each component thoroughly before moving to next
• Documentation During: Write docs alongside code for clarity and completeness
• Production Mindset: Build for production from start, not prototype then rebuild

Technical Best Practices

• Service Worker Global: Register on app load, not just in permission dialog
• Timezone Handling: Store user timezone, convert server-side for accuracy
• Cleanup Logic: Always provide unsubscribe path with data removal
• Test Notifications: Let users verify setup immediately, don't wait for cron

Production Deployment Guide

Prerequisites

1. Generate VAPID keys: npx web-push generate-vapid-keys
2. Set Supabase secrets: VAPID_PUBLIC_KEY and VAPID_PRIVATE_KEY
3. Deploy Edge Functions to Supabase
4. Verify cron job scheduled in database
5. Confirm HTTPS on production domain

Validation Checklist

• [ ] Permission dialog appears and requests access correctly
• [ ] Test notification button sends immediate notification
• [ ] Subscription record created in database
• [ ] Service Worker registered and active
• [ ] Scheduled notification arrives within 30-minute window
• [ ] Disable toggle removes subscription from database
• [ ] Edge Function logs show successful delivery
• [ ] No CORS errors in browser console

Monitoring

• Edge Function Logs: Supabase Dashboard → Functions → Logs
• Database Growth: Track notification_subscriptions table size
• Delivery Rate: Monitor sent vs failed in send-daily-reminders logs
• Subscription Health: Query for expired subscriptions (should auto-cleanup)

This project demonstrates that AI-assisted development, when combined with rigorous testing methodology and strategic engineering, can deliver enterprise-grade results in a fraction of traditional development time. The key is treating AI as an accelerator for implementation while maintaining human oversight for architecture, testing, and quality assurance.