Family meal planner

A small Azure app that helps a family plan the evening meal:

• 🧊 Fridge inventory — track what food is in stock.
• 📲 SMS attendance poll — texts the family each afternoon ("home for dinner? Y/N") and tallies the headcount.
• 🤖 AI recipe suggestions — Azure OpenAI proposes dinners matched to the headcount and what's in the fridge, and can discover new ideas from the web.
• 🖥️ Web dashboard — manage inventory, family, and see/choose tonight's recipe.

Built with TypeScript end-to-end and deployed to Azure. See the full design in .claude/plans/linked-dazzling-naur.md.

Architecture

Concern	Service
Frontend	React + Vite on Azure Static Web Apps
API	Azure Functions v4 (Node) — HTTP, timer, Event Grid triggers
SMS	Azure Communication Services
AI	Azure OpenAI (chat + embeddings)
Database	Azure Database for PostgreSQL Flexible Server + pgvector
Secrets	app settings (move to Key Vault for production)
IaC / deploy	Bicep via azd

mealplanner/
  infra/    Bicep (main.bicep, resources.bicep)
  shared/   shared TypeScript domain types
  api/      Azure Functions (TS)
  web/      React SPA (Vite)
  db/        SQL schema + seed

Prerequisites

This machine does not currently have Node.js. Install:

1. Node.js 20 LTS — https://nodejs.org
2. Azure Functions Core Tools v4 — npm i -g azure-functions-core-tools@4
3. Azure Developer CLI (azd) — https://aka.ms/azd (Azure CLI is already installed)
4. Docker (optional, for a local Postgres) or any local PostgreSQL 15+ with the vector extension.

Local development

npm install                 # installs all workspaces

# 1. Start a local Postgres and apply the schema
#    (example with Docker)
docker run -d --name mp-pg -e POSTGRES_PASSWORD=postgres -p 5432:5432 pgvector/pgvector:pg16
psql postgres://postgres:postgres@localhost:5432/postgres -c "CREATE DATABASE mealplanner"
psql postgres://postgres:postgres@localhost:5432/mealplanner -f db/001_init.sql
psql postgres://postgres:postgres@localhost:5432/mealplanner -f db/002_seed.sql

# 2. Configure the API
cp api/local.settings.json.example api/local.settings.json   # then fill in ACS / OpenAI keys

# 3. Run API + web (two terminals)
npm run dev:api             # func start on :7071
npm run dev:web             # vite on :5173  (proxies /api -> :7071)

Open http://localhost:5173.

Auth: locally the SWA auth headers aren't present, so the API's isAuthenticated check will 401. For local testing either run behind the SWA CLI (swa start http://localhost:5173 --api-location http://localhost:7071), which injects a mock principal, or temporarily relax the auth guard.

Tests

npm run build --workspace shared
npm --workspace api run build && node --test api/dist/test/*.test.js

Covers the SMS reply parser and the fridge-match ranking (the two pure helpers).

Deploy to Azure

azd auth login
azd env new mealplanner
azd env set POSTGRES_ADMIN_PASSWORD "<a-strong-password>"
azd up                      # provisions infra + deploys api & web

After the first deploy, one-off manual steps:

1. Apply the DB schema to the provisioned Postgres (db/001_init.sql).
2. Buy an SMS-enabled phone number in the Communication Services resource and set ACS_FROM_NUMBER on the Function App.
3. Wire inbound SMS → create an Event Grid subscription on the ACS resource for Microsoft.Communication.SMSReceived pointing at the smsInbound function.
4. Configure Entra ID app registration values (AAD_CLIENT_ID, AAD_CLIENT_SECRET) on the Static Web App, and assign the admin role to yourself.

Build phases

Delivered so far: Phase 0 (scaffolding/IaC) + Phase 1–3 core (data model, inventory & family CRUD, SMS attendance poll, AI suggestions, choose recipe). Remaining: richer learning/discovery (Phase 4) and polish (Phase 5) — see the plan file.