OrbitOrca Put my app online

Your coding agent ships it, not you.

Claude Code or Cursor takes your app from your laptop to a live address with sign-in, a database and payments. You answer a few questions in the chat you already use.

Scroll and follow the orca

Right now your app only runs on your laptop.

Stop 1 of 7

Point your agent at your code.

Tell Claude Code or Cursor to put your app on OrbitOrca. It creates the app from your GitHub repo or a ZIP.

Your part: approve the GitHub connection, once.

Claude Code, OrbitOrca MCP server

YouPut my-shop on OrbitOrca.
Claude CodeCreating the app from your GitHub repo.
my-shopcreated from GitHub

YouApprove the GitHub connection. Once.

Stop 2 of 7

One question, then the check.

Before the first deploy your agent asks whether the app already has real users or data. Then we check the code for anything that would break it or leak a key.

A must-fix finding blocks the build until your agent applies the fix. You never open the code.

Readiness check

Claude CodeIs this app already running somewhere with real users or data?
YouNo, only on my laptop.
  • ✓ Start instructions found
  • ✓ No keys exposed to the browser
  • ! A password is in the code ✓ Password moved out of the code
Must fix. Build blocked. Fixed by Claude Code. Ready to build.
Claude CodeTaking the password out of the codeOut of the code now. Check passed.

Stop 3 of 7

Then it ships.

Your agent starts the build, watches it and picks the address. You get an https link anyone can open.

A failed build never touches the running app. A bad version rolls back. A crash restarts itself.

Going live

YouShip it.
  1. Preparing
  2. Starting
  3. Live
https://my-shop.apps.orbitorca.com
Claude CodeIt's up. The name my-shop was free, so that's your address. Logs look clean.

Stop 4 of 7

Sign-in and a database, switched on.

Your agent turns on sign-in. Sign-up, log-in and password reset come ready. Your users live in your app's Postgres database, created on the first deploy.

You see who signed up in the dashboard.

Users and data

YouPeople need accounts.
Claude CodeTurning on sign-in. It starts with the next deploy, and users go in your Postgres database.
  • ✓ Sign-up, log-in and password reset
  • ✓ Postgres database, wired to your app
ana@example.comsigned up just now

Stop 5 of 7

Charge for it. Stripe stays with us.

Your agent turns on payments, creates the product and price, then adds checkout and the paywall to your code from the prompt we give it. Customers pay you, through Stripe.

Your app never holds a Stripe key. It only asks us: has this user paid? Your part: verify your Stripe account.

Payments

YouCharge $12 a month for Pro.
Claude CodeEnabling Stripe, creating Pro at $12 a month, then wiring checkout and the paywall into the code.

Product

$12 a month

Pro

In your app

Pay Has this user paid? Yes

YouVerify your Stripe account: identity and bank.

Stop 6 of 7

Secrets it can set, never read.

Your app needs an OpenAI key? Your agent creates the slot and hands you a link. You type the value in the dashboard, and your app can use it.

No one can read it back afterwards. Not even your agent.

Keys

Claude Codemy-shop needs an OpenAI key. I can't see secrets, so I made the slot. Type the value here.Open the key field

YouType it in the dashboard. Once.

OpenAI keysk-x7Kp2Qm9aL4d••••••••••••
my-shop OpenAI Anthropic Google Maps

Stop 7 of 7

Your own domain, one record.

Add your domain in the dashboard and copy one record to where you bought it. We check it and issue the certificate.

Your first link keeps working underneath.

Your domain

TypeNameValue
A@203.0.113.10Copy

YouAdd this record where you bought the domain.

Checking Connected, certificate issued

https://my-shop.apps.orbitorca.com https://myshop.com

my-shop.apps.orbitorca.com still works

https://myshop.com

Say “ship it”. Charge for it. Keep your code.

Sign-in, a Postgres database, Stripe payments, analytics and your own domain, around the app you already wrote.

Put my app online

GitHub or ZIP. Your code, users and data stay yours.

The backend, switched on by your agent.

Each piece below is something your agent can call through the OrbitOrca MCP server, from Claude Code, Cursor, VS Code, Windsurf or Claude.ai.

Sign-in
Sign-up, log-in and password reset, run for you per app. Users live in your app's database. You see the list in the dashboard.
Database
Postgres, created on the first deploy. Your app gets the address as DATABASE_URL. Shared or your own dedicated one, depending on the plan.
Payments
Stripe Connect through us. Your customers pay you. We hold the keys and who has paid; your app asks us and opens checkout.
Analytics
Visitors, pageviews, top pages and referrers. Your agent switches it on. No code to add.
Address
A subdomain on apps.orbitorca.com straight away. Your own domain with one record; we check it and issue the certificate.
Running
A failed build never touches the running app. A bad version rolls back on its own. A crashed app restarts itself.

Guardrails, because an agent is driving.

Three rules an agent cannot talk its way around.

Readiness check
Before a build we check the Dockerfile, keys left in the code, dependencies and secrets exposed to the browser. A must-fix finding blocks the build and comes with a fix prompt for your agent.
The live question
Before the first check or deploy, your agent has to ask whether the app already runs with real users or data, and record your answer. Until it does, we refuse both. A live app is never restructured by accident.
Write-only secrets
Your agent can create a secret but never passes or sees its value. You type it in the dashboard. No one can read it back, not even your agent.