Deploy to your Cloudflare account

Flary deploys into your Cloudflare account. You own the Worker, Durable Objects, D1 databases, R2 buckets, Queues, secrets, and stored data. You do not manage a VPS or an always-on Node process.

Normal deployment

npx flary deploy

The command performs these steps:

1. Verify Wrangler OAuth and account access.
2. Build the project and generate the final Wrangler configuration.
3. Let Wrangler provision binding-only D1 and R2 resources and named Queues.
4. Upload required secrets from a temporary mode-0600 file.
5. Apply D1 migrations when the project is a dashboard or Mail.
6. Deploy and check /health, plus /setup for dashboard and Mail projects.
7. For a personal backend with a provider, call the example function with its generated bearer token.

For Mail projects, deployment also enables Email Routing and Email Sending for the configured domain. Email Routing replaces the domain’s inbound MX records, so review the Flary Mail guide before the first deployment. Email Sending requires a Workers Paid plan.

The temporary secret file is always removed. Repeated setup and deployment reuse stable Worker and resource names. Setup never deletes a Worker, database, bucket, queue, credential, archive, or thread.

Check a project

npx flary doctor

Doctor checks Node, the package manager, Wrangler authentication, account selection, local required secrets, generated bindings, the dashboard migration file, optional Docker support, and deployed health when a URL is known.

Optional resources

Browser Run is added only when app.browser() is registered. Sandbox is added only when app.sandbox() is registered. Sandbox also needs a running Docker-compatible builder and a paid Workers plan.

If automatic provisioning is not supported by the selected account or plan, Wrangler reports the resource. Create only that resource manually, keep its binding name, and run flary deploy again.