Workflow · Ship
Ship your first app to production
From a repository that runs on localhost to a production environment in your own cloud account, with a domain, TLS and a rollback.
When to use it
You built something that works on your machine and want it live, properly, without becoming an infrastructure engineer first.
- Specialists
- 7
- Stages
- 6
- Tier
- Plan → Act
- Duration
- Typically one sitting
Start it with one prompt.
Paste it into Claude Code, Claude Desktop or Cursor once Kapten is connected. Replace the names with yours; Kapten asks for anything it cannot infer.
Connect your agentDeploy this repo to production on my AWS account in eu-west-3. Use a managed Postgres, put the API behind api.example.com and keep it under 150 €/month.You ask
› Deploy this repo to production on my AWS account in eu-west-3. Use a managed Postgres, put the API behind api.example.com and keep it under 150 €/month.
Stage 1
NavigatorArchitecture
Map the application
Detects the web app, API, worker and the Postgres and Redis they expect. Lists the environment variables the code reads.
Stage 2 · 3 agents in parallel
ShipwrightBuild & containers
Make it buildable
Writes a Dockerfile per service, pins base images and opens a pull request with the changes.
BosunInfrastructure
Plan the infrastructure
Plans the cluster, network, managed Postgres, a bucket and the secrets, tagged and scoped to this environment.
PurserCost
Estimate the bill
Prices the plan line by line against your budget and flags anything that would push past it.
Stage 3
WardenSecurity review
Review exposure
Checks which endpoints are public, that the database is private and that no secret is baked into an image.
Gate 01 · stops for your approval
Merge the Dockerfile pull request
Stage 4
KaptenOrchestration
One plan for your approval
Merges everything into a single plan: resources, releases, cost, findings. Nothing has changed yet.
Gate 02 · stops for your approval
Approve the merged plan (resources, cost, findings)
Stage 5
BosunInfrastructure
Provision
Applies the approved infrastructure plan through the scoped role in your account.
Stage 6
HelmsmanReleases
Release and verify
Rolls out each service behind health checks, attaches the domain and TLS, and confirms it serves traffic.
You get
- A running production environment in your account
- A pull request with build files
- An architecture diagram on the canvas
- A cost estimate and a security report
- A one-step rollback
Related guide
You built it with an AI. Here is how to ship it.
More workflows
Keep the crew busy.
A preview environment per pull request
Give each pull request a complete, isolated copy of the application with seeded data, and remove it when the branch closes.
› Set up preview environments for every pull request on github.com/acme/atlas. Seed Postgres from fixtures, cap each preview at 2 € a day and delete it on merge.
- Navigator
- Bosun
- Purser
- Kapten
- Helmsman
Database migration rehearsal
Run a risky schema migration against a restored copy of production first, measure it, and only then schedule the real one.
› Rehearse migration 0042_split_orders against a copy of production. Tell me how long it locks, and whether the previous release still works after it.
- Bosun
- Helmsman
- Lookout
- Kapten
Install into a customer’s cloud
Deploy your product into an enterprise customer’s own account with a repeatable shape, a scoped role and a clear split of responsibilities.
› Prepare an install of our product in Globex’s AWS account, eu-central-1. They will create the role. Generate the access request and the plan.
- Navigator
- Warden
- Purser
- Bosun
- Helmsman
Build with Kapten
Start with one service.
Build from there.
Tell us what you want to deploy. We review each request and follow up about fit and onboarding on your preferred cloud.
Free during early access. No card required. Cloud usage billed separately.