yourai CLI v0.0.6 · yourai.bdtren.com

yourai deploy [-e] [--dry-run]

Configure how people reach your AI when it should not stay on loopback only. You can enter from yourai init by choosing Deploy to remote (--access remote), or run yourai deploy later from a project folder.

The first question is where yourai should run. Cloudflare is only an optional DNS/HTTPS checklist afterward — not a third place to run the model.

Mode A — build and run on a remote server

You already have a droplet or VPS. The wizard asks for the provider (DigitalOcean or generic VPS) and user@ip, detects or generates an SSH key, prints the public key and the provider security URL, then waits until ssh works. It installs yourai on the remote host if needed, syncs the project (skipping yourai_modules/), builds, binds for remote access with an API key, and starts the server. You open the cloud firewall for the port and use http://SERVER_IP:port.

Mode B — self-host on this PC + remote IP access

yourai stays on this computer. Deploy sets server.host = 0.0.0.0, requires an API key, prints LAN and (if found) public IP URLs, and shows OS firewall / router hints. If you have no usable public IP or cannot port-forward, it can walk you through a tunnel (cloudflared preferred, or Tailscale Funnel).

Commands and files

yourai deploy           # wizard if needed; else run the saved mode
yourai deploy -e        # re-run the wizard (overwrites .yourai/deploy.toml after confirm)
yourai deploy --dry-run # print steps without SSH / start

Settings live in .yourai/deploy.toml (mode = "remote_run" or "local_expose"). Exact browser and firewall steps are copied to .yourai/deploy-README.md. With yourai init --yes --access remote, a stub is written so CI does not hang; finish with yourai deploy -e.

Example — Duy opens the wizard

$ cd ~/study-buddy
$ yourai deploy -e

Where should yourai run?
? Choose one:
> Build and run yourai on a remote server
  Self-host on this PC and grant remote access via an IP

Same Wi‑Fi only? Prefer Local network (--access lan) instead of a full deploy.

Questions or problems? support@bdtren.com · Contact page.