Edit rows. Read the diff. Commit.

A database client that stages every edit as SQL you review.

Show these rows as Rowan psql

shop on db.example.com orders customers Read and write

customers
orders
order_items
products
refunds

status = 'pending'Sort by created

id	customer	total	status	created
1041	Example Kitchen	120.00	pending	09:12
1042	Example Bakery	48.50	~~pending~~ shipped	09:40
1043	Example Florist	75.00	pending	10:05
1044	Example Kitchen	32.00	pending	10:31
1045	Example Cafe	210.00	pending	11:02
1046	Example Florist	~~64.00~~ 60.00	pending	11:47

shop=> SELECT * FROM orders WHERE status = 'pending' ORDER BY created_at;
  id  |    customer     | total  | status  |   created_at
------+-----------------+--------+---------+------------------
 1041 | Example Kitchen | 120.00 | pending | 2026-10-19 09:12
 1042 | Example Bakery  |  48.50 | pending | 2026-10-19 09:40
 1043 | Example Florist |  75.00 | pending | 2026-10-19 10:05
 1044 | Example Kitchen |  32.00 | pending | 2026-10-19 10:31
 1045 | Example Cafe    | 210.00 | pending | 2026-10-19 11:02
 1046 | Example Florist |  64.00 | pending | 2026-10-19 11:47
(6 rows)

shop=> UPDATE orders SET status = 'shipped' WHERE id = 1042;
UPDATE 1
shop=> -- no review, already written

UPDATE orders SET status = 'shipped' WHERE id = 1042; 2 changes staged Review SQL

One window per table

Rows, schema, JSON and keys share one screen.

id	customer_id	total
1041	17	120.00
1042	23	48.50
1043	31	75.00

customers.id: 23
name: Example Bakery
email: orders@example.com
city: Leeds

Follow a foreign key

Click customer_id and the customer opens beside the order.

products.attributes, row 8
{
  "colour": "oat",
  "sizes": ["S", "M", "L"],
  "care": {
    "wash": 30,
    "tumble": false
  }
}

JSON cells open as trees

Fold, search and edit nested values in place.

refu⌘ K

Open table refunds
Filter orders by refund_id
Copy refunds as CSV

Every action on ⌘ K

Tables, filters and exports, a few letters away.

id	`bigint`	primary key
customer_id	`bigint`	references customers
total	`numeric(10,2)`	not null
status	`text`	default 'pending'

The schema stays in view

Types, keys and defaults sit beside the rows they shape.

1. Open it read-only

Paste the connection URL from your app’s config.

Test the connection before you save it
Passwords go to the system keychain

New connection

URLpostgres://app@db.example.com:5432/shop

Nameshop

PasswordSaved in keychain

Read-only

TestConnect

2. Filter without writing SQL

Click a column to sort it; type to filter.

city is Leeds orders over 0 Sorted by name

id	name	email	city	orders
23	Example Bakery	orders@example.com	Leeds	14
17	Example Kitchen	hello@example.com	Leeds	9
40	Example Market	shop@example.com	Leeds	3
52	Example Studio	studio@example.com	Leeds	1

leeds-pending.sql

1SELECT c.name, count(o.id) AS pending
2FROM orders o
3JOIN customers c ON c.id = o.customer_id
4WHERE o.status = 'pending'
5  AND c.city = 'Leeds'
6GROUP BY c.name
7ORDER BY pending DESC;

nametext, customers
notestext, customers

3. Write the SELECT yourself

For the joins and counts a filter cannot reach.

Column names complete from the live schema
Results open as a table you can filter again

4. Commit what you read

psql writes on Enter. Rowan waits for you.

psql, already written

shop=> UPDATE orders SET status = 'shipped'
         WHERE id = 1042;
UPDATE 1
shop=> UPDATE orders SET total = 60.00
         WHERE id = 1046;
UPDATE 1

Rowan, commit.sql, waiting for review

BEGIN;
UPDATE orders SET status = 'shipped' WHERE id = 1042;
UPDATE orders SET total = 60.00 WHERE id = 1046;
COMMIT;

Get the download link

One email, with the installer and nothing else.