Desktop app · macOS, Windows, Linux

A calm database manager for PostgreSQL, MySQL, MariaDB and Redis.

Every engine with its full feature set, not a lowest common denominator. Production is red wherever it appears, destructive statements ask before they run, and your passwords stay in your own keychain.

Download for macOSWindows Linux

Free and open source (AGPL-3.0). Version 0.1.7 Preview.

Search objects, queries, actions⌘K

Revenue by dayordersSession cache

1
2
3
4
5
6
7
8

-- Revenue by day, last 30 days
SELECT date_trunc('day', o.created_at) AS day,
       count(*)                          AS orders,
       sum(o.total)                      AS revenue
FROM   orders o
WHERE  o.created_at >= now() - interval '30 days'
  AND  o.status = 'paid'
GROUP BY 1 ORDER BY 1 DESC;

ResultsMessagesPlanHistory

#dayordersrevenue

12026-10-18 00:00:00+001,204148,220.50

22026-10-17 00:00:00+001,388171,904.10

32026-10-16 00:00:00+001,351162,087.35

42026-10-15 00:00:00+001,290155,410.00

52026-10-14 00:00:00+001,175139,822.75

62026-10-13 00:00:00+00982118,006.40

72026-10-12 00:00:00+001,021121,945.90

82026-10-11 00:00:00+001,412176,330.20

92026-10-10 00:00:00+001,377168,511.05

PRODUCTIONOrders primary · PostgreSQL 17.2 · via SSHAuto commit30 rows · 112 ms

Four engines, each one in full

PostgreSQL13 and newer
MySQL8.0 and newer
MariaDB10.6 and newer
Redis6.2 and newer, Valkey 8

Over TLS and SSH tunnels. Features follow the server version you connect to, so pg_stat_io appears on PostgreSQL 16 and Redis functions on 7.0.

Everything you open a database client for. Free.

The whole app is open source under AGPL-3.0. Every change you make in a designer shows its SQL before it runs.

Query editor and result grid

Real PostgreSQL, MySQL and MariaDB dialects with schema-aware completion. The grid streams millions of rows, edits cells in place and draws EXPLAIN as a plan.

Structure editors

Tables with columns, indexes, constraints and partitions. Views, materialized views, functions, triggers, sequences, types and events, each in its own editor.

Redis tools

A key browser by SCAN pattern, editors for strings, JSON, hashes, lists, sets and sorted sets, streams with consumer groups, pub/sub, a console, memory and diagnostics.

Users and permissions

Planned

PostgreSQL roles, a privilege matrix, default privileges and RLS policies. MySQL and MariaDB accounts and roles. Redis ACL users with a live ACL SETUSER preview and ACL DRYRUN.

Dump, import and export

Planned

SQL dumps from one consistent snapshot, Run SQL file with a dialect-aware splitter, imports from CSV, JSON, Excel, Parquet and XML, exports to nine formats. Memory stays flat on large tables.

Safety around production

Environment tags that color the tree, the tab and the status bar. Read-only connections, confirmation for DROP, TRUNCATE and DELETE without WHERE, manual transactions, and cancel on every engine.

Statistics historyPlanned

Every connect leaves a layer.

Varve will take a snapshot of engine statistics each time you connect, and every 15 minutes while you stay connected. Snapshots are never deleted for you, so today can sit next to your last connect, last quarter, or the first day you connected.

Days you did not connect show as gaps, never as an invented line.
Stored on this device only. You delete history per connection when you choose.
Table growth, cache hits, slow queries, replication and Redis memory.

Today Varve shows live server status, without the history.

orders · total sizeOrders primary · one dot per snapshot

Range shown: 90 days. Options in the app:Last connect30 d90 dAll

Size now: 22.0 GB
Since last connect: +0.31 GB
Cache hit ratio: 99.2 %

Sync and transferPlanned

Compare two databases. Read the script. Then decide.

Each tool will run as a cancellable job with a log, and can write the SQL to a file instead of executing it. Save the settings as a profile and run them again next week.

Structure Sync: Per-object diff in dependency order. Data loss changes need an explicit checkbox.
Data Sync: Inserts, updates and deletes per table, compared by key with chunked hashing on the server.
Data Transfer: Any SQL engine to any SQL engine or a file, with an editable type mapping. Redis to Redis too.

Structure SyncOrders stagingOrders primaryProduction

Included:public.ordersdifferent
Included:public.refund_reasonsonly in source
Not included:public.order_linesdata loss: type narrowing
Not included:public.productsidentical

Source

CREATE TABLE orders (id bigint,status order_status,refund_reason text,created_at timestamptz);

Target

CREATE TABLE orders (id bigint,status order_status, created_at timestamptz);

ALTER TABLE public.orders ADD COLUMN refund_reason text;

Teams and cloud syncPlanned

Share where to connect. Keep how to yourself.

Connection profiles, folders, saved queries and job profiles will sync across your devices and into team workspaces, through a sync server. Host it yourself for free, or use Varve Cloud. Each member connects with their own database user, so the database audit trail stays meaningful.

Passwords, passphrases and SSH keys never leave your device. Not in a personal workspace, not in a team.

Owner, Admin, Editor and Viewer roles
Restricted folders for sensitive connections
Workspace policies as guard rails on production
Revision history for saved queries

Acme team5 members

Kim de VriesOwner
Priya NairAdmin
Jesse VosEditor
Tom BergViewer
[email protected]Invite pending

BillingRestricted to Admins and Jesse

Orders primary is shared. Your password is not.Jesse connects as jesse, with a password stored in this Mac's Keychain.

WrappedPlanned

Your year with your databases, read back in December.

A short story of cards built from your activity log and statistics: top connections, busiest hour, fastest growing table, rows moved, production confirmations. It will be computed on your device and stay private until you export a card.

Exports redact connection, host, database and table names by default, and the preview shows exactly what leaves.

SafetyProduction

Production confirmations: 41.

Times you said no: 3.

Example card · Wrapped 2026

Pricing

The app is free. Only hosting costs money.

Every feature is in one free app. Sync and teams need a server: run it yourself for free, or let us host it as Varve Cloud, for a fee, once it launches.

See what ships today

Varve

Freeopen source, AGPL-3.0

All four engines and their structure editors
Query editor, result grid, data editing, EXPLAIN
Redis tools, live server status, activity log
In-app updates
Planned, also free:
Users and permissions, dump, import and export
Statistics history, Structure Sync, Data Sync, Data Transfer, Wrapped

Download Varve

Sync and teams

Plannedself-hosted free, Varve Cloud paid

Sync between your devices
Team workspaces, roles and policies
Self-hosted: free, on your own server
Varve Cloud, hosted by us: paid, billed per organization (seats)

How hosting works

Open source

Everything is AGPL-3.0, and it stays readable.

Drivers, editors, the grid and every tool live in the public repository. Contributions come in under the Developer Certificate of Origin, so there is no CLA to sign. Features plug in through a small public API that can add commands, jobs and UI slots, never replace them. The planned sync server will live in the same repository, under the same license.

View on GitHub How to contribute

crates/extension/src/lib.rspublic

pub trait Extension: Send + Sync {
    fn id(&self) -> &'static str;   // "acme.hello"
    fn register(&self, reg: &mut Registry);
}

// Registry lets an extension add, never replace:
//  commands, job kinds, migrations,
//  activity actions, hooks, UI slots

Questions

Straight answers.

Do my database passwords go to the cloud?

No. Passwords, SSH passwords, passphrases and key files stay in your system keychain: macOS Keychain, Windows Credential Manager or Secret Service on Linux. The planned cloud sync carries where to connect, never how.

Is Varve really free?

Yes. Varve is one app, free and open source under AGPL-3.0, with every feature. There is no paid edition, no locked menu item and no upgrade prompt.

Then what costs money?

Only Varve Cloud, our managed hosting. Sync and team workspaces will need a server, which is part of the same open-source repository. You can run it yourself for free, or let us host it, billed per organization (seats). It is planned, and the price is not set yet.

Does the cloud sit between me and my database?

Never. Varve connects straight from your machine, over TLS or an SSH tunnel. Query results and statistics are not uploaded, and query history stays on the device. Uploading it will be opt-in.

Do statistics follow me to another computer?

Statistics history is planned, and it will be stored locally, per device. Syncing it is an open question we have not decided yet.

Can workspace policies stop someone from changing production?

Workspace policies are planned as guard rails against mistakes, not as a security boundary. Anyone with valid credentials can use another tool, so real enforcement belongs in database permissions.

Which engines are coming next?

Version 1 covers PostgreSQL, MySQL, MariaDB and Redis. SQLite, SQL Server, Oracle and MongoDB are out of scope for now; the driver layer is built so they can be added later.