Skip to content

🍸 An AI-powered home bar advisor that helps you craft great mocktails or cocktails with whatever bottles you have.

License

Notifications You must be signed in to change notification settings

darth-dodo/cocktail-cache

Repository files navigation

Cocktail Cache

Your cabinet. Your mood. Your perfect drink.

Live Demo

GitHub License: MIT Tests: 751 passing Coverage: 78%


An AI-powered drink advisor that helps you craft great mocktails and cocktails with whatever ingredients you have. Chat with Raja, your AI mixologist, to get personalized drink recommendations, technique guidance, and smart suggestions for your next purchase.

Whether you are mixing spirits or skipping them entirely β€” designated drivers, health-conscious choices, Dry January, or simply in the mood for something refreshing without alcohol, Raja has you covered.


Meet Raja, Your AI Mixologist

Raja Chat Interface

Raja personifies bartender veterans from Bombay. He knows his craft, tells great stories, and speaks with warmth:

"Arrey bhai! Feeling relaxed after a long day? With your bourbon and lemons, let me suggest a Gold Rush β€” ekdum first class for unwinding!"

Just tell Raja what you have and how you're feeling. He'll find the perfect drink.


How It Works

1. Tell Raja What You Have

Add your ingredients to your cabinet β€” spirits, mixers, bitters, juices, syrups, and fresh ingredients. Whether you stock a full bar or just have some citrus and ginger beer, your cabinet is saved automatically.

2. Describe Your Mood

"Celebrating a promotion" β€’ "Quiet evening alone" β€’ "Cocktail nights with friends" β€’ "Sunday afternoon vibes"

3. Get Your Perfect Drink

Raja recommends drinks you can actually make, with step-by-step instructions adapted to your skill level.

4. Know What to Buy Next

Raja tells you which single ingredient will unlock the most new drinks β€” maximum impact for your next purchase.


Features

Feature Description
142 Drinks 103 cocktails + 39 mocktails with detailed recipes
Smart Matching Only shows drinks you can make with your cabinet
Skill Adaptation Beginner-friendly to adventurous techniques
Next Purchase Advice ROI-based suggestions for cabinet expansion
Browse & Search Explore the full catalog with filters
Mobile-First Designed for use in the kitchen

Screenshots

Chat Tab Cabinet Tab Browse Tab

Chat with Raja β€’ Build Your Cabinet β€’ Browse All Drinks


Try It Now

cocktail-cache.onrender.com

No signup required. Just start chatting with Raja.


For Developers

Tech Stack & Architecture

Tech Stack

Component Technology
Runtime Python 3.12
API FastAPI
AI CrewAI + Anthropic Claude
Frontend HTMX + Jinja2 + Tailwind
Deployment Render

Codebase Health

Metric Value
Test Suite 751 tests
Code Coverage 78%
Models 100% covered
Services 96%+ covered
Type Checking Strict mypy
Linting Ruff + pre-commit hooks

Quick Start

git clone https://github.com/darth-dodo/cocktail-cache.git
cd cocktail-cache

cp .env.example .env
# Add your ANTHROPIC_API_KEY

make install
make dev
# Visit http://localhost:8888

Architecture

flowchart LR
    User[User] --> Chat[Chat with Raja]
    Chat --> AI[CrewAI Agents]
    AI --> Claude[Anthropic Claude]
    AI --> Data[(142 Drinks<br/>180 Ingredients)]
    AI --> Recipe[Recipe + Tips]
    Recipe --> User
Loading

AI Agents

  • Raja Bartender β€” Conversational AI with Bombay personality
  • Drink Recommender β€” Matches cabinet + mood to drinks
  • Recipe Writer β€” Generates skill-adapted instructions
  • Bottle Advisor β€” Calculates next bottle ROI

Documentation

Development Commands
Command Description
make install Install dependencies
make dev Start dev server (port 8888)
make test Run test suite
make check Linting and type checks
make format Format code

License

MIT License β€” see LICENSE for details.


Built with
CrewAI β€’ Anthropic Claude β€’ FastAPI β€’ HTMX

Try it Live β†’

About

🍸 An AI-powered home bar advisor that helps you craft great mocktails or cocktails with whatever bottles you have.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •