Overview | Docs | American Cloud

Your AI assistant already writes your software. On American Cloud, it can deploy and operate it too.

These pages are recipes: each one gives you a battle-tested prompt to paste into your assistant — Claude Code, Cursor, or any MCP client — plus a step-by-step account of what the assistant will do with it. The assistant does the work through the American Cloud MCP server; you review and confirm.

[ claude code · deploy with one sentence ]

Cost shown first, you approve, then it builds.

Before you start

Set up the MCP server for your client — Claude Desktop, Claude Code, Cursor, or others.
Deployments create real resources, so you'll need a read-write API key and the --allow-writes flag — see the overview for how both fit together.
For deploy workflows, a coding agent (Claude Code) is the strongest pairing: it combines American Cloud tools with a terminal, so it can provision a server and deploy to it in one conversation.

Every recipe follows the same rule: the assistant shows you a cost estimate before creating anything. You always know the monthly price before a resource exists.

Deployment recipes

Deploy a Next.js app — from working code to a live URL with a domain and TLS
Object storage — S3-compatible storage for uploads, assets, backups, and archives
Kubernetes — provision and scale clusters with MCP tools, operate workloads with kubectl
Deploy a Docker Compose app — if it runs with docker compose up, it can run here
Your own PaaS: Coolify — git-push deploys on a server you own, at a flat monthly price
Self-host Supabase — the full Supabase stack on your own VM
Scale out behind a load balancer — the "after your first VM" sequel
Backups and restore drills — take backups and actually prove they restore

Migration playbooks

Moving from another platform? The assistant inventories what you have, proposes the American Cloud equivalent with a cost estimate, and executes the move in phases — with a DNS cutover at the end, not a leap of faith at the start.

[ claude code · migrate in one conversation ]

Price the equivalent first, build on approval, cut over on your timeline.

Teach your agent the way here

Add one file to your repository and any AI coding agent knows how to deploy that project to American Cloud — conventions, canonical flows, cost-estimate-first behavior included:

The AGENTS.md drop-in

Looking for ideas first?

Things to try with the MCP server collects smaller prompts — audits, cost reviews, incident triage — that work with a read-only key and zero risk.