WhatNow

WhatNow is a multi-app wellness platform that combines fitness coaching, workout analytics, nutrition planning, restaurant discovery, skin and hair analysis, and a community experience into one connected system.

The codebase is organized as several independent Next.js services that share authentication, data, and navigation context across the platform.

Architecture

The repository currently contains six user-facing services:

NowWhat/
|- base/                   # Port 3000 - login, profile, landing page, AI fitness coach
|- skin-hair-analysis/     # Port 3002 - skin/hair profile, image analysis, recommendations
|- nutrition-wellness/     # Port 3003 - pantry planner, dish optimizer, recipes, insights
|- nutrition-yelp/         # Port 3004 - restaurant discovery, health scoring, food scan
|- fitness-dashboard/      # Port 3005 - workout analytics and trends
|- community/              # Port 3006 - posts, events, mood check-ins, people-like-you
|- docs/                   # Project notes, deployment docs, implementation details

High-level flow:

• base acts as the main entry point and navigation hub.
• Each module exposes its own UI and API routes.
• MongoDB stores shared user, session, and feature data.
• Gemini powers generation and image-understanding features where unstructured AI is useful.
• Deterministic backend logic handles validation, filtering, scoring, persistence, and fallbacks.

Main Features

Shared authentication

• Email/password registration and login
• Shared session-token authentication across all services
• Secure httpOnly auth cookie
• Cross-origin auth handoff routes so users can move between services without re-logging in

AI fitness coach

• Voice-first fitness workflow in the base app
• User can request an exercise and generate an analyzer for it
• MediaPipe Pose-based live tracking
• Rep counting, form checks, and spoken feedback
• Session logging to MongoDB

Fitness dashboard

• Session totals and rep totals
• Form, posture, arm-position, and visibility analytics
• Joint symmetry charts
• Daily rep timeline and trend charts
• Mood-linked motivation view

Nutrition wellness

• Nutrition profile with goals, calorie targets, protein targets, preferences, allergies, and restrictions
• Pantry meal generation
• Authentic dish optimization with identity-preserving validation
• Recipe library with save, search, filter, duplicate, and tweak flows
• Custom recipe creation
• Wellness insight cards and session-memory summaries

Restaurants and food scan

• Yelp restaurant search by location, category, price, and sort
• AI-generated restaurant health scoring
• Saved favorites and dining-behavior tracking
• Food image scan for calorie and macro estimates

Skin and hair analysis

• Skin/hair profile management
• Face and scalp image analysis
• Loved products tracking
• Product recommendation generation
• Rule-based wellness insight generation
• Raw uploaded image bytes are discarded after analysis

Community

• Discussion feed with posts, comments, tags, and upvotes
• Wellness events with RSVP
• Mood check-ins and recent mood history
• "People Like You" discovery

Cross-module assistant

• Floating What Now? Agent UI in multiple apps
• Rebuilds a user context summary from recent activity across modules
• Uses backend tooling to verify personal-data questions against MongoDB instead of guessing

Tech Stack

• Frontend: Next.js 14 App Router, React, TypeScript
• Styling: Tailwind CSS
• Database: MongoDB
• Authentication: shared opaque session tokens validated against MongoDB
• AI: Google Gemini API
• Fitness pose tracking: MediaPipe Pose
• Charts and motion: Recharts, Framer Motion
• Voice: Web Speech API, optional ElevenLabs integration
• External APIs and datasets: Yelp Fusion, Open Beauty Facts

Authentication Model

The implemented auth flow is:

1. A user logs in through the base app.
2. The server creates a random opaque session token.
3. That token is stored in the MongoDB sessions collection with an expiry.
4. The browser receives the token in a secure httpOnly auth_token cookie.
5. When the user opens another service, an auth handoff route creates a short-lived handoff grant and sets the same session token cookie on the target origin.
6. Each service validates the cookie by looking up the session in MongoDB.
7. Logout deletes the stored session and clears the cookie.

This gives centralized session invalidation and consistent login state across services.

Local Development

Prerequisites

• Node.js 20+
• npm
• MongoDB connection string
• Gemini API key
• Yelp API key for nutrition-yelp

1. Install dependencies

From the repo root:

npm run setup

2. Create local env files

Copy each app's example env file to .env.local:

• base/.env.local.example
• skin-hair-analysis/.env.local.example
• nutrition-wellness/.env.local.example
• nutrition-yelp/.env.local.example
• fitness-dashboard/.env.local.example
• community/.env.local.example

Common variables used across apps:

MONGODB_URI=...
MONGODB_DB=wellbeing_app
GEMINI_API_KEY=...
NEXT_PUBLIC_BASE_URL=http://localhost:3000
NEXT_PUBLIC_FITNESS_URL=http://localhost:3005
NEXT_PUBLIC_NUTRITION_URL=http://localhost:3003
NEXT_PUBLIC_RESTAURANTS_URL=http://localhost:3004
NEXT_PUBLIC_SKIN_URL=http://localhost:3002
NEXT_PUBLIC_COMMUNITY_URL=http://localhost:3006

Service-specific extras:

• nutrition-yelp: YELP_API_KEY
• base: optional NEXT_PUBLIC_ELEVENLABS_API_KEY
• nutrition-wellness: optional NUTRITION_MEMORY_CRON_SECRET

Note:

• Some env examples still include legacy JWT_SECRET fields from an older auth approach. Those are stale compatibility leftovers and are not the active auth mechanism described above.

3. Start all services

From the repo root:

npm run dev

This starts:

• base on http://localhost:3000
• skin-hair-analysis on http://localhost:3002
• nutrition-wellness on http://localhost:3003
• nutrition-yelp on http://localhost:3004
• fitness-dashboard on http://localhost:3005
• community on http://localhost:3006

Then open:

http://localhost:3000

4. Start services individually

If needed:

cd base && npm run dev
cd skin-hair-analysis && npm run dev
cd nutrition-wellness && npm run dev
cd nutrition-yelp && npm run dev
cd fitness-dashboard && npm run dev
cd community && npm run dev

Data Overview

Important collections used across the platform include:

• users
• sessions
• user_profiles
• fitness_sessions
• fitness_exercise_biomechanics
• nutrition_profiles
• pantry_items
• generated_recipes
• saved_recipes
• recipe_modifications
• nutrition_insight_memory
• nutrition_sessions_summary
• skin_hair_profiles
• skin_logs
• hair_logs
• loved_products
• product_recommendations
• favorites
• clicks
• community_posts
• community_comments
• community_moods
• community_events
• community_connections

Deployment

Each service includes its own deployment assets, including Dockerfiles and env examples under each service's deployment/ directory.

The current deployment docs target containerized deployment, with the broader repository plan documented in:

• DEPLOYMENT.md
• base/deployment/README.md
• community/deployment/README.md
• fitness-dashboard/deployment/README.md
• nutrition-wellness/deployment/README.md
• nutrition-yelp/deployment/README.md
• skin-hair-analysis/deployment/README.md

Documentation

Additional notes live in docs/, including:

• docs/HOW_TO_RUN.md
• docs/SESSION_NOTES.md
• docs/AUTH_IMPLEMENTATION.md
• docs/AUTHENTICATION_SUMMARY.md

Troubleshooting

App-to-app navigation fails

• Check the NEXT_PUBLIC_*_URL variables in every app's .env.local.
• Make sure all apps are running on the expected localhost ports.

Authentication fails across modules

• Confirm the base app can create sessions in MongoDB.
• Confirm the target service can reach the same MongoDB instance.
• Check that cross-app handoff URLs match your local ports.

AI features fail

• Verify GEMINI_API_KEY is present in the app using that feature.
• nutrition-yelp also needs YELP_API_KEY.
• Many AI-backed flows include deterministic or mock fallbacks, but some features still require the upstream key to be present for full functionality.

Build or module issues

• Re-run npm run setup
• Delete stale .next folders if necessary
• Restart the affected service