Jobs & Pipelines is the data workbench inside Manage. Open it from the sidebar and you see every database your sites run on — MariaDB and PostgreSQL alike, both first-class — with their tables, views and columns, sample rows, and a SQL editor to ask your own questions. Queries you want to keep can be saved and are shared with everyone on your account.
What appears
Every site created from the catalogue has its own database: WordPress, WooCommerce, PrestaShop, Ghost, Matomo and Nextcloud run on MariaDB; Directus, Strapi, n8n and Umami run on PostgreSQL; custom PHP and Node.js sites use whichever you attached. All of them are listed in the tree on the left, grouped by site. Expand one to browse its tables; open a table for its columns, the first 50 rows and a description you can write yourself.
Plans
Jobs & Pipelines is included in Pro and Business. On Starter the page still shows your databases, but querying, sample rows and saved queries unlock when the site moves to Pro or Business.
Read-only by design
The editor never changes a site's database. Every query runs as a separate read-only login that only has SELECT rights, inside a read-only transaction, and only SELECT, WITH, SHOW, EXPLAIN, DESCRIBE and VALUES statements are accepted — one at a time. Anything that writes, changes the schema, locks rows, sleeps or reads files is refused before it is queued, and the editor tells you why.
The published limits are the standard offer:
- up to 1,000 rows per query (500 by default — set a higher limit in the editor);
- 30 seconds per query (20 by default);
- 2 queries running at the same time and 2,000 queries a day per account;
- results are kept for 24 hours, the run history for 30 days.
Need more? Contact us — larger row, time and daily limits are available on request, for a fee, on both MariaDB and PostgreSQL databases.
Who can use it
Owners, admins and developers on the account, and anyone invited to a site as a collaborator, can browse and query that site's database. The billing and viewer roles do not see raw rows, so the page is not available to them. Refreshing the cached schema is for owners, admins and developers.
Saved queries
Save a query with Ctrl/Cmd+S and it appears in the rail on the right for everyone on your account. The author, or an account owner or admin, can rename or delete it.
From a script or your AI assistant
The same workbench is available through the API and the MCP server, behind a separate data token scope. It is opt-in: tick it when you create the token — no existing token gains it, and neither does an assistant you connected earlier.
GET /api/v1/data/databases— your databases;GET /api/v1/data/databases/{id}/schema— tables and columns (answers202while it prepares the first time);POST /api/v1/data/query— one read-only statement, waits up to 30 seconds for the rows;GET /api/v1/data/runs/{id}— a run to poll.
An assistant connected over MCP gets three tools: list_databases, get_schema and run_query. Every query is recorded in your audit log with a fingerprint of the SQL, never the text or the rows.
What is coming
Your own warehouse databases with tables you can write to, dashboards built from saved queries, scheduled jobs and pipelines that copy data from a site into the warehouse, and an assistant that drafts queries and descriptions for you. The "Ask AI" and "Generate with AI" buttons are already on the page and switch on when that lands.