Multi-tenant Operations API

Backend-first SaaS reference system that demonstrates tenant isolation as an authorization boundary, not a UI convention.

Capabilities

• GraphQL queries and mutations with cursor pagination, status/search filters and a depth limit.
• Organization-scoped PostgreSQL queries with RBAC, monthly quotas and invitations.
• Access tokens plus hashed, revocable refresh tokens.
• Redis-backed login rate limiting.
• DataLoader for batched createdBy member resolution.
• PostgreSQL-backed background jobs with a separate worker, retries and dead-letter state.
• MongoDB history documents with organization indexes and idempotent event IDs.
• Next.js console using the real GraphQL API for organization summary and operation creation.
• Unit tests, Docker Compose, CI and ADRs describing the security and storage boundaries.

Quick start

Requirements: Node.js 22+, Docker and Docker Compose.

cp .env.example .env
npm install
docker compose up -d postgres redis mongo
npm run dev:api

In separate terminals:

npm run dev:worker
npm run dev:web

Open http://localhost:3200. The local demo account is admin@operations.local / change-me-now; change it before exposing the stack.

GraphQL example

query Dashboard {
  organization { name planKey quotaUsed quotaLimit }
  operations(first: 20, search: "billing") {
    nodes { id name status createdAt createdBy { email role } }
    pageInfo { hasNextPage endCursor }
  }
}

mutation CreateOperation($name: String!) {
  createOperation(name: $name) { id name status createdAt }
}

Send the access token from POST /auth/login as Authorization: Bearer <token> to /graphql.

Validation

npm run lint
npm run typecheck
npm test
npm run build
docker compose config --quiet

After starting the full Compose stack, the tenant boundary and token rotation integration test can be run with:

$env:RUN_INTEGRATION_TESTS = 'true'
npm --workspace @ops/api run test:integration

The API and worker are intentionally separate processes. PostgreSQL is authoritative for tenant and quota decisions; MongoDB is used only for organization history, and Redis is used for rate limiting. See the ADRs.