MealBrain
jtloudon/meal-brain
Sous chef AI que conhece suas receitas, preferências e restrições
Ver no GitHubUm assistente AI para planejamento de refeições que centraliza receitas, listas de compras e planos alimentares, com um sous chef inteligente que entende suas preferências e restrições dietéticas. A IA sugere sem impor, mantendo você sempre no controle das decisões finais.
Instalação
npx convez-ai jtloudon/meal-brain1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Planejar refeições semanais personalizadas
- Encontrar ideias de pratos com ingredientes disponíveis
- Gerar listas de compras automáticas categorizadas
- Gerenciar receitas de múltiplas fontes em um só lugar
Veredito

Solução bem arquitetada que resolve um problema real de organização culinária. O diferencial está em um contexto AI genuíno (não é chatbot genérico) e na filosofia de não sobrescrever decisões do usuário — bom exemplo de IA assistente versus IA impositiva.
README
A tradução para o português deste README ainda está na fila.
MealBrain
Stop juggling recipes (websites, Pinterest, magazines), non-centralized grocery lists, and meal plans on paper. MealBrain is one place for everything, with an AI sous chef that actually knows your recipes, lists, plans, and dietary limitations. Helpful, never bossy.
Why I Built This
I couldn't find what I needed in the App Store: a meal planning app with an AI sous chef that actually understood my preferences and could help me think through weekly meals.
Every app I tried either had no AI, or bolted on a generic chatbot that didn't know anything about my recipes, my dietary rules, or my family's preferences.
So I built MealBrain.
The AI Sous Chef
The core differentiator is an embedded AI assistant with full context about your household:
- Your recipes - ingredients, tags, ratings, notes
- Your meal plans - what's scheduled for the week
- Your grocery lists - what you need to buy
- Your dietary rules - allergies, preferences, restrictions
What It Can Do
Ask it anything:
- "What can I make with the chicken in my fridge?"
- "Plan dinners for this week - we're busy Tuesday and Thursday"
- "Add ingredients for Tacos to my grocery list"
- "What's a quick breakfast I haven't made recently?"
Philosophy: Helpful, Never Bossy
The AI suggests but never surprises. All write operations (adding meals, modifying lists) require your approval first. You stay in control.
Features
- Recipes - Create, import from URLs, import with camera pictures,tag, rate, and organize
- Meal Planner - Calendar-style week view, drag meals around
- Grocery Lists - Auto-categorized by aisle (leverage Claude for undefined mappings), tracks which recipe needs what
- Mobile-First - Cloud deployed, PWA installable on any device, optimized for one-handed use
Tech Stack
| Layer | Technology |
|---|---|
| Frontend | Next.js 14, React, TypeScript, Tailwind CSS |
| Database | Supabase (PostgreSQL + Row Level Security) |
| Auth | Supabase Auth (magic link + password) |
| AI | Anthropic Claude API (Haiku for categorization, Sonnet for chat) |
| Hosting | Vercel (frontend) + Supabase Cloud (database) |
Getting Started
Prerequisites
- Node.js 18+
- Docker (for local Supabase)
- Anthropic API key
Quick Start
# Clone and install
git clone https://github.com/jtloudon/meal-brain.git
cd meal-brain
npm install
# Copy environment template
cp .env.local.example .env.local
# Edit .env.local with your API keys
# Start local Supabase
npx supabase start
# Run migrations
npx supabase db reset
# Start dev server
npm run dev
Environment Variables
Copy .env.local.example and fill in:
| Variable | Description |
|---|---|
NEXT_PUBLIC_SUPABASE_URL | Your Supabase project URL |
NEXT_PUBLIC_SUPABASE_ANON_KEY | Supabase anonymous key |
SUPABASE_SERVICE_ROLE_KEY | Supabase service role key (server-side only) |
ANTHROPIC_API_KEY | Your Claude API key from Anthropic Console |
Database Migrations
The SQL migrations in supabase/migrations/ reflect iterative development - learning and evolving the schema through real-world usage rather than upfront design.
They tell the story of how the app evolved: starting simple, adding features, refactoring as patterns emerged. This is intentional.
Self-Hosting
For cloud deployment (Vercel + Supabase Cloud), see docs/cloud-setup-and-deployment.md.
Estimated cost: ~$2-5/month for a 2-person household
- Supabase: Free tier
- Vercel: Free tier
- Claude API: Pay-per-use (~$2-5/month typical usage)
Documentation
Architecture and design decisions live in /docs:
- Architecture - System design and tech choices
- Data Models - Database schema
- AI Behavior Contract - How the AI operates
- Cloud Setup - Deployment guide
Contributing
Issues and feature ideas welcome. PRs considered - please open an issue first to discuss.
Author
Jesse Loudon - AI, Analytics & Business Insights
Spec-driven development — I defined behavior, constraints, and product decisions; implemented with Claude Code.
License
MIT - see LICENSE