noot 🍎

An AI-powered nutrition logging web app. Just say what you ate!

Demo 📹

demo.mp4

YouTube video demo - In better quality

Available Nutrients

This project aims to track 48+ total nutrients using a centralized definition system:

Adding nutrients is now super easy! See Nutrient System Documentation

See the full list of defined nutrients here.

Quick Add Example:

# Add to config/nutrients.yml
- key: taurine_mg
  type: float64
  unit: mg
  category: amino_acid
  per_100g: true
  original: true
  go_field_name: Taurine
  json_tag: taurine_mg
  display_name: Taurine

Then run: script/generate-nutrients → All 20+ files automatically updated!

Consumption Flow Diagram

This diagram shows the simplified user journey from input to saved meal data, focusing on the user experience and key business logic rather than technical implementation details. The full technical flow is documented separately here.

graph TB
    %% User journey starts - Multiple input methods
    Start([User Input Options:<br/>🎤 Voice Recording<br/>✏️ Text Description<br/>⭐ Duplicate Favorite]) --> Auth{User Authentication<br/>JWT or API Key}
    
    %% Authentication - simplified
    Auth -->|✅ Authenticated User| AuthOK[User Verified<br/>JWT or Pro API Key]
    Auth -->|❌ Not Logged In| AuthFail[Login Required]
    AuthFail --> End([End])
    
    %% Input processing flow
    AuthOK --> InputMethod{Input Method?}
    
    %% Three input paths
    InputMethod -->|🎤 Voice| AudioProcess[Audio Upload & Transcription<br/>Max 50MB, AI converts speech to text]
    InputMethod -->|✏️ Text| TextProcess[Direct Text Processing<br/>Skip transcription step]
    InputMethod -->|⭐ Duplicate| DuplicateProcess[Copy Existing Favorite<br/>Skip AI processing entirely]
    
    %% Error handling
    AudioProcess -->|❌ Invalid File| UploadError[Error: Invalid Audio File]
    TextProcess -->|❌ Empty Text| TextError[Error: Text Required]
    DuplicateProcess -->|❌ Not Found| DuplicateError[Error: Favorite Not Found]
    
    UploadError --> End
    TextError --> End
    DuplicateError --> End
    
    %% Success paths converge
    AudioProcess -->|✅ Transcribed| AIProcess[AI Processing Pipeline]
    TextProcess -->|✅ Text Ready| AIProcess
    DuplicateProcess -->|✅ Copied| SaveDirect[Save Duplicated Meal]
    
    %% AI Processing - simplified into logical steps
    AIProcess --> Step1[Step 1: Extract Food Items<br/>AI identifies individual foods from text]
    Step1 -->|✅ Foods Identified| Step2[Step 2: Get Nutrition Data<br/>Smart lookup with AI assistance]
    Step1 -->|❌ Failed| ProcessError[Error: Could not understand input]
    ProcessError --> End
    
    %% Smart nutrition lookup - business logic
    Step2 -->|✅ Success| NutritionLookup{How do we find nutrition data?}
    Step2 -->|❌ Failed| ProcessError
    
    %% Different data sources - simplified
    NutritionLookup --> Cache[Check Our Database<br/>for Previously Calculated Foods]
    NutritionLookup --> ProductDB[Search External Product Databases<br/>for Branded Items]
    NutritionLookup --> AI[Ask AI for Nutrition<br/>Analysis and Estimates]
    
    Cache -->|Found| UseCache[✅ Use Cached Data<br/>Fast Response]
    Cache -->|Not Found| ProductDB
    
    ProductDB -->|Found Brand Match| UseProduct[✅ Use Product Data as Context<br/>AI gets real product info for<br/>more accurate nutrition calculation]
    ProductDB -->|Not Found| AI
    
    AI --> UseAI[✅ AI Calculates Nutrition<br/>Comprehensive Analysis]
    
    %% All paths lead to nutrition data
    UseCache --> NutritionReady[Nutrition Data Ready<br/>for All Food Items]
    UseProduct --> NutritionReady
    UseAI --> NutritionReady
    
    %% Final steps - user value  
    NutritionReady --> Summary[Calculate Meal Summary<br/>48+ Total Nutrients including<br/>Calories, Protein, Vitamins, Minerals]
    Summary --> Save[Save to User's Meal History<br/>with Labels and Timestamps]
    SaveDirect --> DirectSuccess[Return Duplicated Meal<br/>📱 User sees copied nutrition data]
    
    %% Success response
    Save -->|✅ Saved Successfully| Success[Return Complete Results<br/>📱 User sees comprehensive nutrition data]
    Save -->|⚠️ Save Failed| PartialSuccess[Return Nutrition Data<br/>⚠️ Not saved to history]
    
    Success --> UserActions[User Can:<br/>📊 View on Dashboard<br/>📝 Add to Log<br/>⭐ Mark as Favorite<br/>🏷️ Add Labels]
    PartialSuccess --> UserActions
    DirectSuccess --> UserActions
    UserActions --> End
    
    %% Key benefits callout
    NutritionReady --> Benefits[Key Benefits:<br/>🚀 Fast responses via 4-tier caching<br/>🎯 Real product data guides AI decisions<br/>🤖 AI fills gaps for everything else<br/>📊 48+ comprehensive nutrients tracked<br/>⭐ Easy favorites and duplication<br/>🏷️ Flexible labeling system]
    
    %% Styling for business audience - High contrast for accessibility
    classDef userAction fill:#0D47A1,stroke:#000000,stroke-width:3px,color:#FFFFFF
    classDef success fill:#1B5E20,stroke:#000000,stroke-width:3px,color:#FFFFFF
    classDef error fill:#B71C1C,stroke:#000000,stroke-width:3px,color:#FFFFFF
    classDef process fill:#E65100,stroke:#000000,stroke-width:3px,color:#FFFFFF
    classDef benefit fill:#4A148C,stroke:#000000,stroke-width:3px,color:#FFFFFF
    classDef input fill:#6A1B9A,stroke:#000000,stroke-width:3px,color:#FFFFFF
    
    class Start userAction
    class AuthOK,UseCache,UseProduct,UseAI,Success,PartialSuccess,DirectSuccess,SaveDirect success
    class AuthFail,UploadError,TextError,DuplicateError,ProcessError error
    class AudioProcess,TextProcess,DuplicateProcess,AIProcess,Step1,Step2,Summary,Save process
    class Benefits,UserActions benefit
    class InputMethod input

Loading

Quick Start

Backend (Go API)

1. Setup Dependencies:

   script/bootstrap

2. Get API Keys:

   • OpenAI API key from platform.openai.com

3. Configure:

   # create .env file
   
   # Required
   OPENAI_API_KEY=<key>
   
   # Server Configuration - Use port 3001 for API to avoid conflict with SvelteKit frontend
   PORT=3001
   ENV=development
   
   # CORS Configuration - Allow SvelteKit frontend
   CORS_ALLOWED_ORIGINS=http://localhost:3000
   
   # Logging & Debugging (Enable detailed error reporting)
   LOG_LEVEL=DEBUG
   DEBUG=true
   
   # OpenAI Configuration
   OPENAI_TRANSCRIBE_MODEL=gpt-4o-mini-transcribe
   
   TRANSCRIBE_LANGUAGE=en
   #OPENAI_TRANSCRIBE_RESPONSE_FORMAT=
   
   # Upload Configuration
   MAX_UPLOAD_BYTES=104857600  # Maximum upload size in bytes (default: 100MB)
   
   # Database Configuration - Use Supabase for local and production
   # Follow Supabase local development setup: https://supabase.com/docs/guides/local-development

4. Run the API: PORT=3001 script/server (use --production for production mode).

Frontend (SvelteKit)

1. Setup Dependencies:

   script/bootstrap

2. Run the web app:

   script/frontend

   Or to start without opening the browser:

   script/frontend --no-open

3. Use:

   • Visit http://localhost:3000 (SvelteKit frontend)
   • API runs on http://localhost:3001 (Go backend)
   • Press the microphone button on the record page
   • Say what you consumed (e.g., "I had a latte with organic whole milk and Greek yogurt with blueberries")
   • See your nutrition summary with client-side unit conversion (grams ↔ ounces)

Mobile (Capacitor)

First, in two separate terminal windows, run the backend and frontend:

1. script/server (backend)
2. script/frontend (frontend)

Now, in a third terminal, run:

npm run mobile:sync
npm run ios

API Endpoints

v1 API

• GET /api/v1/health — Health check with server status
• POST /api/v1/consumption — Multipart form with audio field (webm/opus/mp3/wav)
• GET /api/v1/consumptions — View stored consumptions with optional date range filtering. Supports tier-based limits: Free users can access up to 7 days, Pro users up to 365 days (development mode only)
• GET /api/v1/docs — Interactive API documentation via Swagger UI (development mode only)
• GET /api/v1/openapi.yaml — OpenAPI 3.0.3 specification (development mode only)

Frontend Endpoints

• GET /version — Version and build information including commit SHA, build time, and git tag

Note: The embedded static frontend has been removed. The API now runs on port 3001 by default, and the SvelteKit frontend runs on port 3000.

Development

Backend (Go)

• Test: script/test
• Lint: script/lint
• Build: script/build (or script/build --single-target for faster iteration)
• Dev Server: PORT=3001 script/server (use script/server --production for production mode)
• API Docs: Visit http://localhost:3001/api/v1/docs when running in development mode

Frontend (SvelteKit Site)

• Dev Server: script/frontend (runs on port 3000 and opens browser)
• Dev Server (no browser): script/frontend --no-open
• Manual Dev: cd apps/web && npm run dev (alternative manual approach)
• Build: cd apps/web && npm run build
• Type Check: cd apps/web && npm run check
• Generate API Types: cd apps/web && npm run generate:api (regenerates types from OpenAPI spec)

API Development

The API follows OpenAPI 3.0.3 specification:

• Specification: /api/v1/openapi.yaml
• Documentation: /api/v1/docs (Swagger UI, development only)
• Generate Types: script/generate-types (requires oapi-codegen via the go toolchain - vendored in this project)
• Frontend API Types: Generated automatically by openapi-typescript from the OpenAPI spec

Database Management

• Reset: script/db reset — Drop all tables and start fresh
• Seed: Handled by Supabase CLI during local development setup
• Dump: script/db dump — View database contents in human-readable format

Deployment

The application supports automated deployment with built-in version tracking:

• Deploy Frontend: script/deploy frontend — Deploy SvelteKit app to Cloudflare Workers with git metadata
• Deploy Backend: script/deploy backend — Deploy Go API to Railway
• Deploy Both: script/deploy — Deploy both frontend and backend

Version Tracking:

• Each deployment automatically includes commit SHA, build time, and git tag
• Frontend deployments inject git metadata as environment variables during build
• Access version info at /version endpoint (both JSON API and UI page)
• Git metadata is extracted automatically during deployment process

Requirements:

• Frontend: CLOUDFLARE_ACCOUNT_ID and CLOUDFLARE_API_TOKEN environment variables
• Backend: RAILWAY_TOKEN environment variable

Technical Details

Backend

• Supabase/PostgreSQL database with CLI-managed migrations and seeding for consumption storage
• OpenAI gpt-4o-mini-transcribe for speech-to-text
• OpenAI gpt-4o-mini for consumption parsing with complete nutrition data
• Audio uploads are streamed to temporary files to avoid memory spikes
• JWT authentication via Supabase Auth for user management
• CORS middleware for cross-origin requests from frontend

Frontend

• SvelteKit with TypeScript
• DaisyUI with custom "noot" theme for styling
• openapi-fetch for type-safe API calls
• Client-side unit conversion (grams ↔ ounces) with localStorage persistence
• MediaRecorder API for audio capture (webm/opus format preferred)
• Responsive design with mobile-friendly recording interface

Architecture

• Frontend: SvelteKit (http://localhost:3000)
• Backend: Go API server (http://localhost:3001)
• Database: Supabase/PostgreSQL with automatic migrations
• Audio Processing: OpenAI Whisper via API

Theme

• Noot DaisyUI Theme
• Hero Icons
• Lucide Icons