MySQL and PostgreSQL · Windows 10 and 11

A database client that never freezes

Fast, parallel, and built for Windows. Every connection and every query tab runs on its own thread — so the slow SELECT you started five minutes ago never blocks the one you need right now.

Download for Windows Source on GitHub

Free and open source, MIT licensed

Two query tabs open on one connection. The tab bar shows the events tab with a running indicator while the orders tab is in front, and the history pane at the bottom shows the events query still Running at 3s on one row while the orders query on the row below it returned 438 rows in 0.002 sec. — The real application — two query tabs on one connection, one of them still working.

Speed

Fast because nothing waits its turn

Most clients multiplex every tab onto one connection, so the work queues up behind whatever is slowest. This one gives each tab its own connection and does the waiting off the UI thread entirely.

The query history pane. Row 6, marked with an amber dot, reads Running at 3 seconds for a join across the events table. Row 7, one second later and marked complete, reads 438 rows returned in 0.002 seconds. — The history pane from that screenshot, full size. Row 6 is **still running at 3s**; row 7 started after it and was **back in 0.002 sec**. Same connection, same moment — no queue.

Threads, not queues

A worker thread per connection and a separate database connection per query tab. Two tabs on one server genuinely run at the same time.

Virtualized grids

Results are virtualized, so a big result set scrolls at the same speed as a small one. Rows are parsed off the UI thread before they ever reach the grid.

Instant schema tree

The schema list is cached on disk per connection, so the tree is populated the moment a connection opens instead of after a round trip.

Real cancellation

A runaway query is killed on the server from a second connection — by thread id on MySQL, by backend pid on Postgres — not just abandoned in the UI.

A look around

The rest of the app

Every screenshot on this page is the real application — no mockups, no illustrations. Click any of them to see it full size.

The main window: schema tree on the left, a syntax-highlighted SELECT in the editor, a result grid of 438 rows, and the query history along the bottom. — **Query and results**Schema tree, editor and grid, with every statement you have run kept in the history pane underneath.

A PostgreSQL 16.14 connection open in a second connection tab, browsing the reporting schema and a daily_revenue table. — **PostgreSQL, side by side**A Postgres server in one connection tab and MySQL in another, each with its own catalogue and its own SQL.

The table designer showing the orders table: a column list with data types and primary key, not-null and auto-increment flags, a detail panel for the selected column, and tabs for Columns, Indexes and Foreign Keys. — **Table designer**Columns, indexes and foreign keys as a form — and Apply opens the `ALTER TABLE` in a new tab instead of running it.

The connections home screen with six saved connections as cards, badges marking the Postgres, SSH and AWS IAM ones, and a production connection tinted red. — **Connections**Saved connections as cards, grouped and colour-coded — the red tint marks the ones you want to think twice about.

Everyday work

Built for the things you actually do all day

Reading schemas, editing a few rows, altering a table, moving data between servers — with the SQL always visible before it runs.

Schema-aware SQL editor

CodeMirror 6 with completion that knows your tables and columns, plus search and formatting.

Editable result grid

Virtualized for large results. Edit cells, review the generated SQL, then apply or revert.

Table designer

Create and alter tables through a form that shows you the exact DDL before anything runs.

Import and export

Drives mysqldump and mysql, or pg_dump, pg_restore and psql, with the command shown in full.

SSH tunnels and IAM

Connect over TCP, through an SSH tunnel, or with AWS IAM auth tokens — refreshed automatically before the 15-minute expiry.

Sessions survive restarts

Open tabs, their result grids, schema cache and layout are written per tab and restored the next time you open the connection.

Offline

The schema is still there when the server isn't

Open a connection with the database unreachable — on a plane, off the VPN, or with the tunnel down — and the schema tree, table lists and columns are all still browsable from cache, alongside every tab you had open.

Right-click any saved connection and choose Open Offline. The cache is filled the first time you open that connection online.

Connection context menu Open Connection Open Offline (cached schemas and tabs) ───────────────────────────── Edit Connection… Duplicate Connection ───────────────────────────── Delete Connection…

	MySQL	PostgreSQL
Schema model	One flat level	Schemas inside one database
Switch schema	`USE`	`SET search_path`
Dump and restore	`mysqldump`, `mysql`	`pg_dump`, `pg_restore`, `psql`
Cancel a query	By thread id	By backend pid

Two engines

MySQL and PostgreSQL, both first class

Not one engine ported onto the other. Each gets its own catalogue queries, its own DDL generation in the table designer, its own dump and restore tooling, and its own way of cancelling a query that someone else is running.

MySQL · MariaDB PostgreSQL

Safety

It never changes your database behind your back

Ever worried about changing the wrong database? Tick Confirm on modifying query when you set the connection up, and every INSERT, UPDATE, DELETE and DDL statement you run from the editor — and every grid edit you apply — stops behind this, with the exact SQL in front of you.

The whole application washed red behind a red confirmation dialog titled: warning, modifying query on analytics — production. It reads: this will change data or schema on this connection, read it carefully before running. Below it the exact UPDATE statement is shown in full, with Cancel and Run modifying query buttons. — The connection is named in the title, the statement is shown in full, and **Cancel** is what has keyboard focus.

Drop, truncate, alter and grid-apply do not execute when you click them. They open a new tab pre-loaded with the SQL, or show you a copyable confirmation — you run it. The only statements that execute on their own are the ones you typed and ran yourself, plus USE and the explicit "select rows, limit 1000" button.

Connection passwords and SSH secrets are encrypted at rest with Windows DPAPI, and updates install quietly in the background when you next close the app.