Cellar Assistant

Uncork the potential of your collection with Cellar Assistant

Features

• TODO

Technology Stack

• Frontend: Next.js
• Backend: Nhost (serverless backend platform with GraphQL API, authentication, and storage)

Getting Started (local dev)

1. Prerequisites:
   • Node and Docker
2. Clone the repository:

   git clone https://github.com/your-username/cellar-assistant

3. Install dependencies:

   cd cellar-assistant
   nvm use
   npm ci

4. Create a Nhost project and set up environment variables:
   • Follow the documentation to install the Nhost cli
   • Create a .env.local file in the project root and add the following variables:

      NEXT_PUBLIC_NHOST_REGION
      NEXT_PUBLIC_NHOST_SUBDOMAIN=local
     

   • Create a .secrets file in the project root and add the following variables:

      # OAuth Configuration
      DISCORD_OAUTH_CLIENT_ID = 'FIXME'
      DISCORD_OAUTH_CLIENT_SECRET = 'FIXME'
      FACEBOOK_OAUTH_CLIENT_ID = 'FIXME'
      FACEBOOK_OAUTH_CLIENT_SECRET = 'FIXME'
      GOOGLE_OAUTH_CLIENT_ID = 'FIXME'
      GOOGLE_OAUTH_CLIENT_SECRET = 'FIXME'
     
      # Nhost Configuration
      HASURA_GRAPHQL_ADMIN_SECRET = 'nhost-admin-secret'
      HASURA_GRAPHQL_JWT_SECRET = '0f987876650b4a085e64594fae9219e7781b17506bec02489ad061fba8cb22db'
      NHOST_WEBHOOK_SECRET = 'nhost-webhook-secret'
     
      # AI Provider Configuration (choose one setup)
      
      # Option 1: Local Development with Ollama (Recommended)
      AI_PROVIDER = 'ollama'
      OLLAMA_ENDPOINT = 'http://localhost:11434'
      OLLAMA_MODEL = 'llama3.2-vision:11b'
      
      # Option 2: Google AI (Simple Cloud Setup)
      # AI_PROVIDER = 'google-ai'
      # GOOGLE_AI_API_KEY = 'your-google-ai-api-key'
      # GOOGLE_AI_MODEL = 'gemini-2.0-flash-exp'
      
      # Option 3: Vertex AI (Enterprise/Production)
      # AI_PROVIDER = 'vertex-ai'
      # GOOGLE_GCP_PROJECT_ID = 'your-gcp-project-id'
      # GOOGLE_GCP_VERTEX_AI_LOCATION = 'us-central1'
      # VERTEX_AI_MODEL = 'gemini-2.0-flash-001'
      # CREDENTIALS_GCP_ID = 'your-credentials-db-id'
     
      # Legacy/Other Configuration
      GRAFANA_ADMIN_PASSWORD = 'FIXME'
      PLACEHOLDER_API_URL = 'FIXME'
     

5. Set up AI Provider (Ollama for local development):

   # Install Ollama (visit https://ollama.ai for installation instructions)
   # Pull the vision model for item analysis
   ollama pull llama3.2-vision:11b
   
   # Start Ollama (usually starts automatically after installation)
   ollama serve

6. Start the local nhost environment:

   nhost up --apply-seeds

7. Start the development server:

   npm run dev

8. Urls
   • Cellar Assistant: http://localhost:3000
   • Postgres: postgres://postgres:postgres@localhost:5432/local
   • Hasura: https://local.hasura.nhost.run
   • GraphQL: https://local.graphql.nhost.run
   • Auth: https://local.auth.nhost.run
   • Storage: https://local.storage.nhost.run
   • Functions: https://local.functions.nhost.run
   • Dashboard: https://local.dashboard.nhost.run
   • Mailhog: https://local.mailhog.nhost.run
9. Logins
   • test@test.com:123456789
   • test2@test.com:123456789

AI Provider Options

Cellar Assistant uses AI for analyzing wine, beer, spirit, and coffee labels from images. You can choose from three AI providers:

🏠 Ollama (Recommended for Development)

• Pros: Free, runs locally, no API keys needed, complete privacy
• Cons: Requires local installation and model download (~7GB)
• Best for: Local development, privacy-sensitive use cases
• Setup: Install from ollama.ai, then ollama pull llama3.2-vision:11b

☁️ Google AI (Simple Cloud Setup)

• Pros: Easy setup with just an API key, fast response times
• Cons: Costs money per request, data sent to Google
• Best for: Simple cloud deployments, prototyping
• Setup: Get API key from Google AI Studio

🏢 Vertex AI (Enterprise/Production)

• Pros: Enterprise-grade, integrates with existing GCP infrastructure
• Cons: More complex setup, requires GCP project
• Best for: Production deployments, enterprise environments
• Setup: Requires GCP project with Vertex AI enabled

Contributing

We welcome contributions from the community! Please feel free to open issues or pull requests.