GenCourse

AI‑powered course generation with a modern web UI, real‑time notifications, and a modular services architecture.

System Diagram

flowchart LR
  subgraph Browser
    A[Web UI\nVite/React] -- HTTP --> G[[Oathkeeper Gateway]]
    A -- WS --> NWS[(Notifier WS)]
  end

  G --> MS[Management Service\nNode TS • Hono • Drizzle]
  MS <--> PG[(PostgreSQL)]
  MS -- AMQP --> RMQ[(RabbitMQ)]

  subgraph Workers
    LW[LLM Worker\nPython] -- consumes --> Q1((course.create))
    LW -- produces --> Q2((course.created))
    LW -->|notifications| QN((notifications))
  end

  RMQ <-.-> Q1
  RMQ <-.-> Q2
  RMQ <-.-> QN

  Q2 --> MS
  QN --> NS[Notifier Service\nGo]
  NS -.-> NWS

  subgraph Auth
    K[Ory Kratos] -. session mgmt .-> G
  end

Loading

Services

• Web UI (Vite + React)

  • Location: web-ui/
  • Dev: npm run dev (port 3030)
  • Env:
    • VITE_PUBLIC_MANAGEMENT_API_URL default http://localhost:4455/api/management
    • VITE_PUBLIC_NOTIFIER_WS_URL default ws://localhost:4455/api/notifier/ws
    • VITE_PUBLIC_ORY_KRATOS_URL default http://localhost:4433

• Management Service (Node, TypeScript, Hono)

  • Location: services/management-service/
  • Dev: npm install && npm run dev (port 3000)
  • Build: npm run build; Start: npm start
  • DB: Postgres
  • Queues:
    • Publishes course.create
    • Consumes course.created and publishes notifications
  • Health: GET /health

• Notifier Service (Go, WebSocket hub)

  • Location: services/notifier-service/
  • Env:
    • AMQP_URL (default amqp://guest:guest@localhost:5672)
    • NOTIFIER_QUEUE (default notifications)
    • HTTP_ADDR (default :8080)
    • WS_PATH (default /ws)
    • HEALTH_PATH (default /healthz)
  • Endpoints:
    • GET /ws (WebSocket)
    • GET /healthz

• LLM Worker (Python)

  • Location: services/llm-worker/
  • Runs as a consumer for course.create, emits course.created, and may emit notifications.
  • Env:
    • AMQP_HOST, AMQP_PORT, AMQP_USER, AMQP_PASS
    • QUEUE_CREATE (default course.create)
    • QUEUE_CREATED (default course.created)
    • GENERATION_MAX_RETRIES, GENERATION_RETRY_DELAY_SECONDS
    • LLM_PROVIDER, LLM_BASE_URL for model selection

• Gateway and Auth (Docker‑only in local dev)

  • Oathkeeper: public gateway at 4455 (admin 4456)
  • Kratos: public 4433 (admin 4434)

Project Structure

• web-ui/ – React app
• services/management-service/ – Course and lesson API
• services/notifier-service/ – WebSocket notifications
• services/llm-worker/ – Course generation worker
• config/ – Gateway and auth configuration
• docker-compose.yml – Local infra and services

Local Development

1. Start backend stack

docker compose up -d

Components:

• Postgres 5432
• RabbitMQ 5672 (UI: 15672)
• Kratos 4433/4434
• Oathkeeper 4455/4456
• Management, Notifier, LLM Worker

2. Start Web UI

cd web-ui
npm install
npm run dev

Open http://localhost:3030

Endpoints

• Gateway base: http://localhost:4455
  • Management API: /api/management
  • Notifier WS: /api/notifier/ws

Contributing

• Follow UI_UX_ACCESSIBILITY.md.
• Keep changes small and focused; add tests where feasible.

License

MIT