SkyMate MVP

Minimal flight booking prototype optimized for Vercel deployment.

Stack

• Next.js 15 (App Router, Node runtime)
• Prisma + Postgres (Vercel Postgres/Neon)
• Zod for validation
• OpenAI (optional, for assistant later)

Getting Started

1. Install deps

pnpm install

2. Run Postgres (no external account needed)

Option A — Docker (recommended):

docker compose up -d

This starts Postgres 16 on port 5432 with DB skymate and password postgres.

Option B — Local Postgres:

Create a database named skymate and ensure it is accessible on localhost:5432.

3. Configure env

Copy .env.example to .env and set DATABASE_URL (Neon/Vercel Postgres). For development you can use a local Postgres too.

For Docker/local Postgres, use:

DATABASE_URL="postgresql://postgres:postgres@localhost:5432/skymate?schema=public"

4. Prisma

pnpm prisma:generate
pnpm prisma:migrate --name init

5. Run dev server

pnpm dev

API

• POST /api/flights/search → { origin, destination, departDate, adults }

• GET /api/flights/offers/:id

• POST /api/bookings (requires x-user-id header and optional Idempotency-Key)

• POST /api/bookings/:id/cancel (requires x-user-id)

• GET /api/bookings?status=...&cursor=... (requires x-user-id)

• GET /api/bookings/:id (requires x-user-id)

• PATCH /api/bookings/:id (requires x-user-id) → update contact and/or passengers

• GET /api/users/me (requires x-user-id or uid cookie)

• PATCH /api/users/me (requires x-user-id or uid cookie)

• GET /api/users/me/bookings?status=...&cursor=... (requires x-user-id or uid cookie`)

• POST /api/chat → Proxies to local Ollama (llama3.2) to answer site questions.

  • Tools: get_user_profile, list_user_bookings, get_next_flight, search_flights

For auth, this prototype uses an x-user-id header. Replace with NextAuth when ready.

Deploy to Vercel

• Add Vercel Postgres integration (or Neon) and set env vars:
  • DATABASE_URL, POSTGRES_PRISMA_URL, SHADOW_DATABASE_URL
• Add a build step that runs:
  • pnpm prisma:generate and pnpm prisma:deploy

Assistant (Local Ollama)

• Install Ollama and pull a model (e.g., ollama pull llama3.2).
• Start Ollama locally (defaults to http://localhost:11434).
• Optional envs:
  • LLM_PROVIDER (default ollama; set to openai to use OpenAI)
  • LLM_MODEL (override default model for selected provider)
  • OPENAI_API_KEY (required when LLM_PROVIDER=openai)
  • OLLAMA_BASE_URL (default http://localhost:11434)
  • OLLAMA_MODEL (default llama3.2)
• Open /assistant in the app to chat.

Switching providers

• Ollama (default): ensure local server is running; set nothing or LLM_PROVIDER=ollama.
• OpenAI: set LLM_PROVIDER=openai and OPENAI_API_KEY. Optionally set LLM_MODEL (e.g., gpt-4o-mini).