managarten/apps/nutriphi/CLAUDE.md

NutriPhi Project Guide

Overview

NutriPhi is an AI-powered nutrition tracking app that allows users to photograph their meals and receive instant nutritional analysis. It uses Google Gemini for image analysis and provides personalized recommendations.

App	Port	URL
Backend	3023	http://localhost:3023
Web App	5180	http://localhost:5180
Landing Page	4323	http://localhost:4323

Project Structure

apps/nutriphi/
├── apps/
│   ├── backend/      # Hono/Bun compute server (@nutriphi/server)
│   │   └── src/
│   │       ├── main.ts
│   │       ├── app.module.ts
│   │       ├── db/           # Drizzle schemas
│   │       │   ├── schema/index.ts
│   │       │   └── db.ts
│   │       ├── meal/         # Meal CRUD
│   │       ├── goals/        # User goals
│   │       ├── favorites/    # Favorite meals
│   │       ├── analysis/     # Gemini AI integration
│   │       ├── stats/        # Daily/weekly statistics
│   │       ├── recommendations/  # AI hints & coaching
│   │       └── health/
│   │
│   ├── web/          # SvelteKit web application (@nutriphi/web)
│   │   └── src/
│   │       ├── lib/
│   │       │   ├── api/client.ts
│   │       │   ├── stores/
│   │       │   │   ├── auth.svelte.ts
│   │       │   │   └── meals.svelte.ts
│   │       │   └── components/
│   │       │       ├── Header.svelte
│   │       │       ├── DailySummary.svelte
│   │       │       ├── MealList.svelte
│   │       │       ├── AddMealButton.svelte
│   │       │       └── ProgressRing.svelte
│   │       └── routes/
│   │           ├── +layout.svelte
│   │           ├── +page.svelte      # Dashboard
│   │           ├── login/+page.svelte
│   │           └── add/+page.svelte  # Photo/text input
│   │
│   └── landing/      # Astro marketing page (@nutriphi/landing)
│
├── packages/
│   └── shared/       # Shared types, utils, constants (@nutriphi/shared)
│       └── src/
│           ├── types/index.ts
│           ├── constants/index.ts
│           └── utils/index.ts
│
├── package.json
└── CLAUDE.md

Commands

Root Level (from monorepo root)

# Start all apps
pnpm nutriphi:dev

# Individual apps
pnpm dev:nutriphi:server     # Backend (port 3015)
pnpm dev:nutriphi:web         # Web app (port 5180)
pnpm dev:nutriphi:landing     # Landing page (port 4323)
pnpm dev:nutriphi:app         # Web + backend together

# Database
pnpm nutriphi:db:push         # Push schema to database
pnpm nutriphi:db:studio       # Open Drizzle Studio

Backend (apps/nutriphi/apps/backend)

pnpm dev                      # Start with hot reload
pnpm build                    # Build for production
pnpm db:push                  # Push schema to database
pnpm db:studio                # Open Drizzle Studio

Web App (apps/nutriphi/apps/web)

pnpm dev                      # Start dev server (port 5180)
pnpm build                    # Build for production

Landing Page (apps/nutriphi/apps/landing)

pnpm dev                      # Start dev server (port 4323)
pnpm build                    # Build for production

Technology Stack

Layer	Technology
Backend	Hono + Bun, Drizzle ORM, PostgreSQL
AI	Google Gemini 2.5 Flash
Web	SvelteKit 2.x, Svelte 5 (runes mode), Tailwind CSS 4
Landing	Astro 5.x, Tailwind CSS
Auth	Mana Core Auth (JWT)

Architecture

Core Features

1. Photo Analysis - Take a photo, Gemini identifies foods and calculates nutrition
2. Text Input - Alternative: describe your meal in text
3. Full Nutrition - Calories, macros, vitamins, minerals
4. Daily Goals - Set and track calorie/macro targets
5. AI Coaching - Personalized tips based on eating patterns
6. Favorites - Save frequently eaten meals
7. Privacy-First - Photos are never stored, only analysis results

API Endpoints

Health

Endpoint	Method	Description
/api/v1/health	GET	Health check

Analysis

Endpoint	Method	Description
/api/v1/analysis/photo	POST	Analyze photo (Base64)
/api/v1/analysis/text	POST	Analyze text description

Meals

Endpoint	Method	Description
/api/v1/meals	GET	List meals (query by date)
/api/v1/meals	POST	Create meal
/api/v1/meals/:id	GET	Get meal details
/api/v1/meals/:id	PATCH	Update meal
/api/v1/meals/:id	DELETE	Delete meal

Goals

Endpoint	Method	Description
/api/v1/goals	GET	Get user goals
/api/v1/goals	POST	Set/update goals
/api/v1/goals	DELETE	Delete goals

Favorites

Endpoint	Method	Description
/api/v1/favorites	GET	List favorites
/api/v1/favorites	POST	Create favorite
/api/v1/favorites/:id/use	POST	Increment usage count
/api/v1/favorites/:id	DELETE	Delete favorite

Stats

Endpoint	Method	Description
/api/v1/stats/daily	GET	Daily summary
/api/v1/stats/weekly	GET	Weekly stats

Recommendations

Endpoint	Method	Description
/api/v1/recommendations	GET	List active recommendations
/api/v1/recommendations/:id/dismiss	POST	Dismiss recommendation

Database Schema

user_goals

Column	Type	Description
id	UUID	Primary key
user_id	UUID	User ID
daily_calories	INTEGER	Daily calorie target
daily_protein	INTEGER	Protein target (g)
daily_carbs	INTEGER	Carbs target (g)
daily_fat	INTEGER	Fat target (g)

meals

Column	Type	Description
id	UUID	Primary key
user_id	UUID	User ID
date	TIMESTAMP	Meal date/time
meal_type	VARCHAR	breakfast/lunch/dinner/snack
input_type	VARCHAR	photo/text
description	TEXT	AI-generated description
confidence	REAL	AI confidence (0-1)

meal_nutrition

Column	Type	Description
id	UUID	Primary key
meal_id	UUID	FK to meals
calories	REAL	Calories (kcal)
protein	REAL	Protein (g)
carbohydrates	REAL	Carbs (g)
fat	REAL	Fat (g)
fiber	REAL	Fiber (g)
sugar	REAL	Sugar (g)
vitamin_*	REAL	Various vitamins
calcium, iron, etc.	REAL	Minerals

favorite_meals

Column	Type	Description
id	UUID	Primary key
user_id	UUID	User ID
name	VARCHAR	Favorite name
nutrition	JSONB	Cached nutrition data
usage_count	INTEGER	Times used

recommendations

Column	Type	Description
id	UUID	Primary key
user_id	UUID	User ID
type	VARCHAR	hint/coaching
message	TEXT	Recommendation text
dismissed	BOOLEAN	User dismissed

Environment Variables

Backend (.env)

NODE_ENV=development
PORT=3023
DATABASE_URL=postgresql://manacore:devpassword@localhost:5432/nutriphi
MANA_CORE_AUTH_URL=http://localhost:3001
CORS_ORIGINS=http://localhost:5180,http://localhost:4323

# Gemini AI (uses gemini-2.5-flash model)
GEMINI_API_KEY=your-gemini-api-key

Note: Get your API key from https://aistudio.google.com/apikey

Web (.env)

PUBLIC_BACKEND_URL=http://localhost:3023
PUBLIC_MANA_CORE_AUTH_URL=http://localhost:3001

Shared Package (@nutriphi/shared)

Types:

• UserGoals - Daily nutrition targets
• Meal, MealNutrition - Meal data
• FavoriteMeal - Saved favorites
• DailySummary, WeeklyStats - Statistics
• AIAnalysisResult - Gemini response format
• Recommendation - AI hints/coaching

Constants:

• DEFAULT_DAILY_VALUES - Reference daily values
• MEAL_TYPE_LABELS - Localized meal names
• NUTRIENT_INFO - Labels, units, colors
• CREDIT_COSTS - Credit pricing

Utils:

• calculateProgress() - Progress towards goals
• sumNutrition() - Sum multiple meals
• formatNutrient() - Display formatting
• detectDeficiencies() - Find nutrient gaps
• suggestMealType() - Based on time of day

Quick Start

1. Create Database

# PostgreSQL must be running
docker compose -f docker-compose.dev.yml up -d postgres

# Create database
PGPASSWORD=devpassword psql -h localhost -U manacore -d postgres -c "CREATE DATABASE nutriphi;"

# Push schema
pnpm nutriphi:db:push

2. Set Gemini API Key

Add to .env.development:

GEMINI_API_KEY=your-gemini-api-key

3. Start Apps

# Backend + Web together
pnpm dev:nutriphi:app

# Or individually:
pnpm dev:nutriphi:server  # Terminal 1
pnpm dev:nutriphi:web      # Terminal 2
pnpm dev:nutriphi:landing  # Terminal 3

4. Open URLs

• Web App: http://localhost:5180
• Landing: http://localhost:4323
• API Health: http://localhost:3023/api/v1/health

Testing API

# Health Check
curl http://localhost:3023/api/v1/health

# Login (get token)
TOKEN=$(curl -s -X POST http://localhost:3001/api/v1/auth/login \
  -H "Content-Type: application/json" \
  -d '{"email": "test@example.com", "password": "password"}' | jq -r '.accessToken')

# Analyze text
curl -X POST http://localhost:3023/api/v1/analysis/text \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"description": "Spaghetti Bolognese mit Parmesan"}'

# Get daily summary
curl http://localhost:3023/api/v1/stats/daily \
  -H "Authorization: Bearer $TOKEN"

Credit System

Action	Credits
Photo Analysis	5
Text Analysis	2
AI Coaching	10

Privacy Features

• Photos are NEVER stored on servers
• Photos are sent directly to Gemini, analyzed, then discarded
• Only nutrition results are saved
• Full data export available (GDPR)
• One-click account deletion

Color Theme

Color	Value	Usage
Primary	#22C55E	Main actions, progress
Secondary	#F97316	Accent, warnings
Accent	#14B8A6	Highlights
Calories	#F59E0B	Calorie displays
Protein	#EF4444	Protein displays
Carbs	#3B82F6	Carb displays
Fat	#8B5CF6	Fat displays