Look inside Postgres. Skip the IDE.

PG Compass is a small desktop app for the thing you do most with a database: open it and see what's there. Browse schemas, read rows, check an index, run a quick query, and get back to work.

Download the latest release View on GitHub

Free and MIT licensed. No account, no telemetry. For Windows and Linux.

Page through a table 25 to 100 rows at a time. Every column shows its type under its name.

Navigation

Schemas first. Every table one click away.

Connections, schemas, tables and views live in one tree. Save a connection with a URI or separate fields, give it a label and a colour, and star the ones you open every day.

Tabs keep their connection's colour, so a production tab never looks like a staging one.

Column	Type	Nullable

Data

Rows or documents. Your call.

Page through a table 25 to 100 rows at a time and filter with a plain WHERE clause. When a table is mostly JSONB, switch to card view and read each row as a document.

Double-click a cell to edit it, or insert, update and delete rows with a preview first.

Rows 240 shop.orders

id int4	customer text	status order_status	total numeric	placed_at timestamptz
1	Ada Okafor	delivered	61.13	2026-09-26 18:30
2	Hugo Okafor	shipped	98.26	2026-09-26 17:03
3	Cleo Lindqvist	paid	135.39	2026-09-26 15:36
4	Jonas Lindqvist	delivered	172.52	2026-09-26 14:09
5	Elif Moreau	pending	209.65	2026-09-26 12:42
6	Greta Tanaka	delivered	246.78	2026-09-26 11:15
7	Ines Reyes	refunded	283.91	2026-09-26 09:48

Query

Ask in SQL. Read the answer.

Every table has a Query tab with a highlighted SQL editor. Results land in the same grid, with the same pagination, card view and export.

It only runs SELECT, so a stray DELETE stays in your clipboard.

Production / shop / orders Query


          1
          SELECT c.name, c.plan, sum(o.total) AS spent
        
          2
          FROM shop.orders o
        
          3
          JOIN shop.customers c ON c.id = o.customer_id
        
          4
          WHERE o.status <> 'refunded'
        
          5
          GROUP BY c.id
        
          6
          ORDER BY spent DESC
        
          7
          LIMIT 4;

Only SELECT statements are allowed. Ctrl+Enter

Press Run query or Ctrl+Enter to see the results.

name	plan	spent
Cleo Reyes	member	1185.92
Jonas Whitlock	free	1173.64
Greta Lindqvist	crew	1161.36
Ada Whitlock	member	1137.80

Appearance

Dark by default. Light when the sun is out.

Both themes are built from the same translucent layers, so panels, tables and dialogs read the same in either. Follow your system if you'd rather not choose, and pick compact or comfortable rows to suit your screen.

Drag the handle, or focus it and use the arrow keys.

Also in the box

The small things you reach for every day.

Six views per table.

Data, Structure, Indexes, Constraints, Triggers and Types sit side by side in one tab.

Export and import.

Export a table or a query's results as CSV or JSON, or dump it as SQL. Import rows from CSV or JSON.

Extension-aware.

pgvector embeddings and PostGIS geometry render as values you can read, not binary blobs.

Read-only when it counts.

Turn on read-only mode and edits disappear from the UI. Write requests are refused outright.

SSL and SSH tunnels.

Connect through a bastion host, or verify the server with your own CA certificate.

Keyboard friendly.

Switch tabs, refresh a view, search the sidebar and run a query without reaching for the mouse.

Download

Get the latest release.

Builds are published to GitHub releases for Windows and Linux, straight from CI. There's no macOS build yet.

Latest release

From GitHub

Version: Checking…
Published: —
Files: —

Read the release notes

Before you get started.

The questions people ask first.

Does it replace pgAdmin or DataGrip?

No, and it isn't trying to. PG Compass covers browsing, reading, light editing and quick queries. Migrations, schema design, stored procedures and backups are out of scope, and they'll stay that way.

Can it change my data?

Only when you ask it to. You can edit a cell, insert rows, and run bulk updates or deletes, which show a preview of the affected rows first. The Query tab only runs SELECT. Turn on read-only mode and the app refuses every write.

Where are my connection details stored?

On your machine. Passwords are encrypted with your operating system's credential store when one is available, and they never leave your computer except to reach your database.

Does it collect any data?

No. There's no account, no analytics and no telemetry. Apart from the databases you connect to, the only network traffic is an update check against GitHub every few hours. It carries nothing about your databases, and you can turn it off in Settings.

Which platforms are supported?

Windows, and Linux through an AppImage or a .deb package. Windows installs update themselves, and on Linux the app tells you when a new version is out. There's no macOS build yet.