You open a table to check one value. That value turns out to be a JSON document: nested objects, arrays and a number long enough to make you look twice. The edit might be tiny. Understanding exactly what you’re changing takes more care.
That’s the kind of work Rowberry is being built for.
I’m building Rowberry, a free, open-source PostgreSQL desktop app that brings tables, SQL and JSON into one local-first workspace. It’s coming soon, with the first release focused on Apple-silicon Macs.
JSON deserves more than a cramped cell
A single JSON column can contain an entire world of application data. Finding the right value shouldn’t mean wrestling with a wall of text.
Rowberry gives you space to explore that structure, search for values and inspect the underlying JSON. Its editing workflow preserves large numbers and original text, because opening a document should never quietly change a value you didn’t touch.
That precision matters when the long number is an identifier rather than something you can afford to round.
The important part happens before Save
Editing database data deserves a little friction in the right places.
Rowberry lets you preview changes before committing them. Table-row editing is tied to a stable row identity, and optimistic conflict checks help detect when someone else has changed the row since you loaded it.
Local draft recovery gives interrupted work another chance. Recovered changes remain drafts until you explicitly apply them; restoring your workspace shouldn’t silently write to your database.
Arbitrary SQL results remain read-only. That’s a deliberate boundary: displaying a result doesn’t necessarily mean there’s a safe, unambiguous way to edit it.
The everyday tools are here too
The JSON focus sits alongside the things you need to get through normal PostgreSQL work: saved connections, schema navigation, filtering, sorting and paged table browsing.
There are SQL tabs with table and column completion, saved queries, cancellation and explicit transaction controls. You can move from browsing a table to writing a query without leaving the workspace.
Those controls still sit on top of PostgreSQL’s own permissions and behaviour. Cancelling a query, for example, isn’t proof that a write has been rolled back.
Your database. Your machine.
Rowberry connects directly to PostgreSQL and keeps its workspace locally. There’s no Rowberry account to create, no hosted service in the middle, no telemetry and no paid tier.
Remembered passwords use the operating system’s credential store when available. Local-first doesn’t make sensitive data harmless, though: saved queries and recovered drafts still deserve the same care as the data they contain.
Coming soon
The core workflows are in place, and the work now is getting Rowberry ready for a public release. Signing, installation checks and release qualification still need to be completed before it’s ready to ship.
The first version has a focused scope: PostgreSQL, Apple-silicon Macs, and a dependable path from finding your data to understanding it and making a deliberate change.
If that sounds like your kind of tool, take a look at Rowberry on GitHub. The source is available now; the public release is coming soon.
Cover illustration generated with AI.
