replica-backend wires up the server side of a clone from the architecture plan. It covers sign up and sessions, database migrations with access rules, Stripe Checkout and webhooks, transactional email, background jobs and third-party integrations.
You create the accounts and keys yourself and the agent writes only variable names into .env.example. Everything runs in test mode until deploy.
What You Get
- Auth with email verification, password reset, optional OAuth, roles and real account deletion
- Checked-in migrations with row level security or an authorisation function on every table
- Stripe Checkout and Customer Portal with signature-verified, idempotent webhook handlers
- Transactional email, retried background jobs and an
.env.examplewith names only - A
replica/backend.mdwith the security checklist ticked
Common Use Cases
- Adding login, teams and roles to a clone that runs on seed data
- Connecting Stripe in test mode with webhooks that survive retries
- Starting Google OAuth verification early for calendar or other sensitive scopes
What Makes It Unique
It sets hard rules about who holds the keys: the user creates every account and live key, the agent never types a password or card number, and only official public APIs with your own credentials are used. It also tests access rules with a second user and treats one-click cancelling as a requirement.
Under The Hood
The skill is instructions only. The agent writes migration files, server code, a seed script and markdown notes into your project, and the checklist mentions running a dependency audit with npm. It reads and writes files and runs commands in your repository. It points at third-party services such as Stripe, Supabase and Resend, using keys you place in env files yourself.