Vending Machine

Vending Machine is a full-stack application based off of Project Vend by Anthropic. It allows for purchasing virtual items offered on the vending machine through an AI Agent, where item selection and restocking is also orchestrated through other AI Agents.

Stack

• Backend: Python, FastAPI, LangGraph, LangChain, and OpenAI
• Frontend: Angular, TypeScript, and Vitest

Repository Layout

.
|-- backend/     FastAPI service, agent graph, prompts, and backend tests
|-- frontend/    Angular application and frontend tests

Prerequisites

• Python 3.10 or newer
• Node.js with npm
• An OpenAI API key for the backend

Backend Setup

From the repository root, run:

cd backend
python -m venv venv
venv\Scripts\python.exe -m pip install -r requirements.txt
copy .env.example .env

Set the backend values in backend/.env. The supported settings are:

• AI_PROVIDER
• AI_API_KEY
• AI_MODEL
• AI_TEMPERATURE
• AI_MAX_TOKENS
• VENDING_CURRENCY

Start the backend from the backend directory:

set PYTHONPATH=src
venv\Scripts\python.exe -m uvicorn vending_machine.main:app --reload

The service runs at http://localhost:8000. FastAPI's interactive API documentation is available at http://localhost:8000/docs.

Frontend Setup

From the repository root, run:

cd frontend
npm install
npm start

The Angular development server runs at http://localhost:4200.

Run the backend and frontend in separate terminals during development.

API Overview

• GET / returns a service health response.
• POST /conversations creates a conversation.
• POST /conversations/{conversation_id}/messages streams assistant output as server-sent events.
• GET /inventory returns every slot in the vending machine.

Testing And Builds

Run backend tests from backend:

set PYTHONPATH=src
venv\Scripts\python.exe -m unittest discover -s tests

Run frontend tests and a production build from frontend:

npm test
npm run build

Do not commit backend/.env or any other secret values.