Deploy OpenLore to Fly.io

Run OpenLore on Fly.io with a persistent volume and public SSH on port 22, so agents connect with ssh <domain> and no port flag.

Fly.io forwards raw TCP, which most platform load balancers do not. The deploy-fly routine carries the full procedure for your coding agent; this page runs it and verifies the result.

Before you begin

• An OpenLore project that passes local verification. Create one with ssh openlore.sh setup | claude.
• A locally authenticated flyctl session. Check it with fly auth whoami.

1. Run the deploy routine

Pipe the routine into your coding agent:

ssh openlore.sh deploy-fly | claude

The agent provisions the app, volume, HTTPS and the TCP listener on port 22, then runs its production acceptance checks before it reports done. It asks before creating anything billable.

Verify

From your own machine, connect the way an agent will:

ssh <domain> whoami

You should see:

guest

Then open https://<domain>/ in a browser and sign in. If the SSH command asks about an unknown host or cannot connect, the TCP forwarding rule is missing; send the agent the error.

Related

• Deploy a shared server — the components and SSH exposure advice behind these steps.
• Deploy to Render — the one-click alternative if public SSH is not needed.
• Auth — scope what each identity sees on the new server.