> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pylonsync.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Pylon Cloud

> Hosted Pylon at www.pylonsync.com/dashboard.

Pylon is fully self-hostable, but if you'd rather not run it yourself, **Pylon Cloud** at [www.pylonsync.com](https://www.pylonsync.com) hosts the same app you'd run on a VPS. Connect a GitHub repo, push to your default branch, and the project is live at `https://your-app.pyln.dev`.

## When to use it

| You should use Cloud if...                                                 | You should self-host if...                                         |
| -------------------------------------------------------------------------- | ------------------------------------------------------------------ |
| You want to ship today, not learn ops                                      | You have strict data residency / on-prem requirements              |
| You don't want to manage Postgres backups, TLS, scaling                    | You already operate Kubernetes / Nomad / your own VPS fleet        |
| You'd rather pay a flat platform fee + per-machine compute than per-server | Your workload spikes hard enough that reserved capacity is cheaper |
| You want managed magic-link email + OAuth credentials out of the box       | You need to run an air-gapped deployment                           |

Both targets run the same binary. You can start on Cloud and migrate to self-hosted (or back) without touching client code — only the `baseUrl` changes.

## Sign up

1. Visit [www.pylonsync.com/dashboard](https://www.pylonsync.com/dashboard) and create an account.
2. Create a project. The project gets a stable URL like `https://your-app.pyln.dev`.
3. Add a payment method (free Hobby tier covers small projects; see pricing below).

## Deploy

There are two ways to deploy to Cloud:

### 1. Push to GitHub (recommended)

Install the Pylon Cloud GitHub App on your repo from the project settings page. Every push to your project's default branch triggers a build + deploy. Pull-request pushes get a preview environment that tears down on merge.

```sh theme={null}
git push origin main
# → Cloud's webhook fires
# → build runs on the platform
# → schema migrates, traffic cuts over
# → live at https://your-app.pyln.dev
```

You can also trigger a manual deploy of any branch from the project's dashboard (useful for redeploying without a fresh push, or shipping a feature branch to the project's machine for testing).

### 2. Manual deploy from the dashboard

Project page → **Deploy** → pick a branch. The control plane resolves the branch head against your installed GitHub App and runs the same build pipeline as a webhook deploy.

### Custom domains

From the project's **Domains** tab, add `app.example.com`. Cloud provisions a TLS certificate (Let's Encrypt) automatically. Add the displayed `CNAME` record to your DNS, wait for verification, and you're live on the custom domain. Both the apex `*.pyln.dev` URL and your custom domain stay valid.

### Environment variables

Project page → **Settings** → **Environment**. Set per-project secrets without committing them to git. They show up in your TypeScript functions as `process.env.STRIPE_SECRET_KEY`, are encrypted at rest, and never appear in logs.

### Hand off to a coding agent

Every org page in Pylon Cloud has a **Hand off to your coding agent** card. Click it, paste the generated prompt into Claude Code (or Codex / OpenCode / Cursor / Aider / grok build), and the agent installs the CLI, signs in, loads the Pylon skill, and asks what you want to build. The session token is exchanged via a 5-minute single-use code — your `pk.*` API key never lands in the chat history. See [Agent handoff](/operations/agent-handoff) for the full flow.

### Pair with Vercel for the frontend

The canonical Pylon production stack is **frontend on Vercel**, **backend on Pylon Cloud**. Vercel handles your Next.js / SvelteKit / etc. build + global edge; Pylon Cloud owns the realtime backend, database, and auth.

Set `PYLON_TARGET=https://your-app.pyln.dev` (your project's Cloud URL) in Vercel's project env vars and the Next.js template's rewrite forwards `/api/*` to your Pylon backend — the browser talks same-origin, session cookies ride along, no CORS needed.

Full checklist: [Deploying to Vercel](/operations/vercel).

## What's included

* **Same binary as `pylon dev`** — no runtime hooks, no vendor proxies.
* **Postgres or SQLite** — choose at project creation. SQLite ships on a Fly volume; Postgres is provisioned as a managed sibling service.
* **TLS** — automatic Let's Encrypt certificates on `*.pyln.dev` and your custom domains.
* **Magic-link email** — managed transactional sender; no SendGrid / Resend account required.
* **OAuth** — bring your own Google / GitHub / etc. OAuth credentials.
* **WebSocket + SSE + shard ports** — exposed automatically; clients connect to the public URL.
* **File storage** — S3-backed, sized by plan.
* **Studio** — every project has the inspector available at `https://your-app.pyln.dev/studio`.
* **Logs + metrics** — request logs, function traces, latency / error charts in the dashboard.
* **On-demand backups** — Fly volume snapshots from the project's **Backups** tab. Pro and Team plans get more frequent retention.

## Scaling

Pylon Cloud runs on Fly Machines under the hood. Every dashboard knob maps to a real infrastructure call:

* **Machine size** — `shared-cpu-1x · 512 MB` baseline; resize up to 4 GB on Pro, 8 GB on Team.
* **Replicas** — 1–3 on Pro, 1–10 on Team.
* **Volume size** — up to 100 GB on Pro, 500 GB on Team.
* **Autostop** — locked on for free tier (machine sleeps when idle, wakes on next request). Toggleable on Pro+ if you need always-warm WebSockets.
* **Multi-region** — pick from any of 13 regions:

| Region                    | Code  |
| ------------------------- | ----- |
| US East (Ashburn)         | `iad` |
| US Central (Chicago)      | `ord` |
| US West (San Jose)        | `sjc` |
| US West (Los Angeles)     | `lax` |
| US Northwest (Seattle)    | `sea` |
| EU (London)               | `lhr` |
| EU (Frankfurt)            | `fra` |
| EU (Amsterdam)            | `ams` |
| Australia (Sydney)        | `syd` |
| Asia (Singapore)          | `sin` |
| Asia (Tokyo)              | `nrt` |
| South America (São Paulo) | `gru` |
| Canada (Toronto)          | `yyz` |

Pricing is per-machine — multi-region equals one machine per region.

## Pricing

Each tier has a flat platform fee plus per-machine compute. Hard quotas (requests, CPU-hours, egress) cap monthly usage; projects pause at the quota until the calendar month rolls over or you upgrade.

### Hobby — \$0

Locked free shape:

* 1 project per organization
* `shared-cpu-1x · 512 MB · 3 GB` volume, 1 replica, autostop forced on
* 100k requests / month
* 1 CPU-hour / month
* 5 GB egress / month
* Single region, SQLite only
* Community support

### Pro — \$20 / org / month

Includes one baseline 512 MB machine, 10 GB volume, 50 GB egress, all features (custom domains, OIDC SSO, audit log, autostop toggle).

* Up to 10 projects per organization, 5 organizations per user
* Quotas: 5M requests, 50 CPU-hours, 250 GB egress per month
* Per-machine pricing:
  * 1 GB always-warm: \$14 / month
  * 2 GB always-warm: \$25 / month
  * 4 GB always-warm: \$50 / month (max)
* Each replica bills at the machine rate × count
* Volume above 10 GB: \$0.25 / GB-month
* Egress above 250 GB: \$0.06 / GB

### Team — \$99 / org / month

Includes one baseline 1 GB machine, 50 GB volume, 250 GB egress.

* Up to 50 projects per organization, 20 organizations per user
* Quotas: 50M requests, 500 CPU-hours, 2 TB egress per month
* Per-machine pricing:
  * 1 GB always-warm: \$12 / month
  * 2 GB always-warm: \$22 / month
  * 4 GB always-warm: \$42 / month
  * 8 GB always-warm: \$110 / month (max)
* Volume above 50 GB: \$0.20 / GB-month
* Egress above 2 TB: \$0.04 / GB
* Multi-region replication, SAML SSO, audit log, role-based access

### Enterprise

Bespoke quotas, single-tenant Fly org or BYOC (AWS, GCP), custom regions, SLA + on-call escalation, migration assistance. Email `cloud@pylonsync.com`.

Live pricing on the [pricing page](https://www.pylonsync.com/#pricing).

## Migrating from self-hosted

`pylon backup` and `pylon restore` (in the framework CLI) snapshot the SQLite database + uploads on a local Fly volume. To migrate into Cloud:

```sh theme={null}
# 1. Snapshot your local data
pylon backup ./snapshot/

# 2. Create the project in the dashboard at www.pylonsync.com/dashboard,
#    with the SAME schema you've been running locally.

# 3. Connect your GitHub repo and push — Cloud builds + deploys the
#    same binary you'd run locally.

# 4. Use the dashboard's "Restore from snapshot" flow on the
#    project's Backups tab to upload ./snapshot/.
```

Then point your clients at the new `baseUrl` (your project's `pyln.dev` URL or custom domain). Sessions are preserved — the session store is part of the snapshot — so users stay signed in.

## Migrating to self-hosted

```sh theme={null}
# 1. Snapshot from Cloud — Backups tab → "Download snapshot"
#    or trigger a backup first with the "Snapshot now" button.

# 2. Restore into your own deployment locally or on a VPS.
pylon restore ./snapshot/
```

No lock-in. The binary is the same; the data format is the same.

## Troubleshooting

**Push didn't trigger a deploy** — Confirm the Pylon Cloud GitHub App is installed on the repo (Project Settings → GitHub) and the push is to the configured default branch. The Deployments tab shows webhook receipt; if nothing arrived, the webhook never fired.

**Custom domain stuck on "verifying"** — DNS propagation can take up to 24 hours. `dig CNAME app.example.com` should show your `*.pyln.dev` target.

**Magic-link emails not arriving** — Check the spam folder for the first send (Cloud uses a shared sender domain that warms up over time). For production, configure your own sender domain in **Settings → Email** to authenticate with SPF / DKIM / DMARC under your domain.

**WebSocket disconnects** — Cloud's load balancer holds connections for up to 4 hours; the sync engine handles reconnects automatically with full-jitter backoff. If you see frequent drops, check **Connections** in the project dashboard.

**Hit a monthly quota** — The project pauses at the limit until the calendar month rolls over (UTC) or you upgrade the organization's plan. The dashboard shows current usage on the project's overview page.

## Status

[status.pylonsync.com](https://status.pylonsync.com) shows live region health and incident history.