# Export and backup

Source: [Export and backup](https://typeroll.com/docs/guides/export-and-backup/)

> Download a site's complete content as JSON, mirror its media, build it yourself with the open-source renderer, and back up a whole self-hosted installation.

Your content stays yours. Every site can be exported at any time, including during a read-only period after a Cloud trial, and the export builds with the open-source renderer without Typeroll Cloud. A self-hosted installation also has an encrypted backup of everything.

## What the content export contains

The export is a zip file with the site’s content as JSON:

* Site settings, Pages, Content types, Page templates and partials.
* Forms, redirects, block types and the media records.
* `export-manifest.json` with the export’s format version, the site and version, the time, and a flat list of every media URL, including image variants.

The files use the same layout the open-source site renderer builds from, under `organizations/<organization>/sites/<site>/`.

Not included:

* **Media files themselves.** They stay in your media storage; the manifest lists their addresses so you can download them.
* **Form submissions, API keys and integration secrets.**
* **Revision history, users and Organization settings.**

A generated Git repository used for publishing is not a backup of the CMS.

## Download an export

* **Portal:** **Site settings → Advanced → Export your content → Download content export**.

* **API:** with a key that has admin permission on the site:

  ```bash
  curl -H "Authorization: Bearer $TYPEROLL_API_KEY" -o site-export.zip \
    https://app.typeroll.com/api/v1/sites/<siteId>/export
  ```

  Add `?version=<id>` to export a branch as it would be published. For a self-hosted portal, use its address instead of `app.typeroll.com`.

Exports need admin permission: editors and Organizations a site is shared with at read or write level cannot download one.

## Mirror the media

The manifest’s media list is enough for a simple download script, for example:

```bash
unzip site-export.zip -d site-export
mkdir -p media && cd media
node -e "for (const url of require('../site-export/export-manifest.json').media_urls) console.log(url)" \
  | xargs -n 1 curl -fsSLO
```

`curl -O` names each file after the last part of its address. Keep the manifest with the files so each one can be matched to its original address.

## Build the site from an export

The export is a working local site for the open-source renderer in [Typeroll/typeroll](https://github.com/typeroll/typeroll):

```bash
git clone https://github.com/typeroll/typeroll
cd typeroll
npm ci
TYPEROLL_FIXTURES_DIR=/path/to/site-export \
TYPEROLL_ORG_ID=<organization> TYPEROLL_SITE_ID=<site> \
TYPEROLL_SITE_URL=https://www.example.com \
npm run build --workspace=@typeroll/site-template
```

The result is a static site in `packages/site-template/dist`. Images keep pointing at your media storage until you mirror them and change their addresses.

## Regular backups

* **Typeroll Cloud:** schedule the export API, for example nightly, and store the zip files and mirrored media where your other backups live. The published website and its media already live in your own Cloudflare account.
* **Self-hosted:** the operator’s backup covers every site, user and media file in one encrypted snapshot, with a verified restore. See [Backups and disaster recovery](https://typeroll.com/docs/guides/self-hosting/#backups-and-disaster-recovery). Test a restore at least quarterly.

## Leaving Typeroll Cloud

Published static sites stay on your hosting account when Cloud access ends; see [Cloud trial and access](https://typeroll.com/docs/guides/cloud-trial/). Download the export during the read and export period, then build it yourself as above or recreate it in another Typeroll installation.

There is no import of an export zip into another site through the portal or API. A self-hosted operator can place the exported files in the installation’s data store, or an AI agent can recreate the content through the API.

## Check it

* Open `export-manifest.json` and confirm the site, version and export time.
* Count the Pages in the export against the site’s page list.
* Build the export locally and open a few pages.