Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/salaboy/pizza-vibe itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 26, 2026Pizza Vibe
Pizza Vibe is an agentic pizza store that uses AI agents to orchestrate the entire pizza ordering workflow — from taking your order through a chat interface, to cooking your pizza and delivering it to your door.
Customers interact with a natural language chat assistant that coordinates behind the scenes with specialized AI agents for store management, cooking, and delivery.
An agents dashboard provides real-time visibility into what each agent is doing, along with the status of ovens, bikes, inventory, and drinks stock.
Getting Started
Prerequisites
- kind (installed automatically via Homebrew if missing)
- kubectl
- Helm
- Docker
- ANTHROPIC_API_KEY (required) — an Anthropic API key for the AI agents
- DASH0_AUTH_TOKEN (optional) — a Dash0 token for cloud-based observability
Setup
export ANTHROPIC_API_KEY=<your-anthropic-api-key>
# Optional: enable Dash0 observability
export DASH0_AUTH_TOKEN=<your-dash0-token>
./scripts/setup-kind.sh
This script creates a KIND Kubernetes cluster, installs PostgreSQL, the observability stack (OpenTelemetry + Jaeger), builds all service images, and deploys the full application.
Accessing the Application
kubectl port-forward svc/store 8080:8080
# Open http://localhost:8080/chat.html to access the chat interface
# Open http://localhost:8080/agents-dash.html to access the agents dashboard
# Jaeger tracing UI
kubectl port-forward svc/jaeger-query 16686
# Open http://localhost:16686
Architecture
The application follows a multi-agent architecture where a React/Next.js frontend communicates with a Go backend (Store Service) via HTTP and WebSocket. The Store Service coordinates three AI agents (built with Java, Quarkus, and LangChain4j) that use tools exposed by Go microservices through an MCP server.
Protocols and Tools
| Technology | Role |
|---|---|
| A2A (Agent-to-Agent) | Communication protocol between the Store Service and the AI agents |
| MCP (Model Context Protocol) | Protocol used by agents to discover and invoke tools provided by the Go services |
| OpenTelemetry | Distributed tracing and observability across all services (exported to Jaeger and optionally Dash0) |
Services
| Service | Language | Description |
|---|---|---|
| store | Go | Central API gateway and orchestrator. Serves the frontend, manages orders, coordinates agents via A2A, and provides real-time updates over WebSocket |
| inventory | Go | Manages pizza ingredient stock and handles ingredient acquisition for pizza preparation |
| oven | Go | Manages pizza ovens, tracks availability, and handles oven reservations for cooking |
| bikes | Go | Manages a fleet of delivery bikes and handles bike reservations for deliveries |
| drinks-stock | Go | Manages beverage inventory, stock levels, and drink acquisition |
| store-mgmt-agent | Java (Quarkus) | AI agent that orchestrates the full order workflow, coordinating between cooking and delivery agents |
| cooking-agent | Java (Quarkus) | AI agent that handles pizza preparation — acquires ingredients, reserves ovens, and manages cooking |
| delivery-agent | Java (Quarkus) | AI agent that handles delivery logistics — reserves bikes and manages the delivery process |
| pizza-mcp | Java (Quarkus) | MCP server that exposes inventory, oven, bike, and drinks-stock tools for agents to use |
| front-end | React/Next.js | Web UI with a chat interface for ordering pizzas and a dashboard for monitoring agents and resources |