Start Pro Trial
48 Releases

Release History

Track all versions, updates, and improvements to VisuaLeaf.

v1.0.2714

Latest
November 7, 2026

VisuaLeaf v1.0.2714 Release Notes

Release Date: November 6, 2026

Covers everything since v1.0.2686.


Headline: Make the Home Page Yours, See a Sync Before It Runs, and Shortcuts That Always Work

The biggest change in this release is the first thing you see: the home page is now yours to arrange. Add pages, drag and resize tiles, pick a preset, and put live dashboards, single charts and your favorite tables right where you open the app.

Also shipping:

  • A dry run for every Compare sync. See exactly what a sync would insert, update and delete, with live progress and a Stop button, and download the full plan as .sql or .js before anything is written.
  • Keyboard shortcuts that always work. Clicking a toolbar button no longer switches off your table shortcuts, SQL tables get the same editing keys as MongoDB, and the Keyboard Shortcuts window tells you which keys are free.
  • Charts that draw themselves. A new chart from a query opens with a sensible chart already on screen, and the field boxes are named after what they control (X axis, Y axis, Slices…).
  • Select several cells and drag them into the query builder. Each cell becomes the right filter for its type, in both SQL and MongoDB.
  • Oracle Table Management opens tables in about 0.2 seconds, down from 4–9 seconds.
  • Connection errors that don't lock you out. A wrong password on Test Connection is now one login attempt, not nine.

Customizable Home Page

The home page used to be a fixed layout. Now you can shape it around how you work: a page for your daily dashboards, another for the tables you open every morning.

What you can see

  • Pages. Name your own tabs on the home page, each with its own tiles. The tab strip only shows up once you have a second page, and VisuaLeaf reopens the page you used last.
  • Live dashboards and charts on the home page. Any saved dashboard, or a single saved chart, can be a tile. A chart tile is drawn exactly as it looks on a dashboard and grows when you make the tile taller. Each tile loads only when you scroll to it, waits politely while its connections are closed (with a Connect & show button), and fills in as soon as they open.
  • Favorites. Right-click any collection or table and choose Add to Home Favorites to open it in one click from the home page.
  • Tiles for every tool, including Query Profiler, Schema Manager, Compare Data, Saved Items, Data Transfer, New Connection and Connection Manager, plus heading tiles you can label yourself.
  • Presets. Classic, Compact, Original (the first VisuaLeaf home page), Sections, Sidebar, Focus, Dashboard first and Calm, each previewed as a thumbnail before you apply it.
  • Three tile looks: Compact, Card and Preview, in ten accent colors. Tint all recolors a whole page at once. Tiles adapt to their size: short tiles become a single line, tall ones show the feature's picture.

What you can do

  • Click Customize, then drag, resize, remove and add tiles back. Your layout is saved automatically.
  • Turn off Snap together on a page to place tiles freely, with gaps where you want them. Tidy up packs the page again.
  • Move several tiles at once. Shift/Ctrl-click or drag a box around tiles to select them, then drag one and the rest follow. The preview shows where every tile will land before you let go. Resizing one selected tile resizes them all.
  • Change color, look or remove a whole selection in one step.
  • Work in two windows. A change in one window shows up in the other right away, and an older window can no longer overwrite your newer layout.
  • Pick a chart for a Chart tile, then Refresh it in place or Open it as a tab. A closed connection, a deleted or unshared chart, or a plan without charts each gets a clear message instead of a blank tile.
  • Ask the AI assistant to do it. Through the MCP connection, an assistant can create pages, apply presets, and add dashboards, charts, tiles and favorites for you.

Keyboard Shortcuts

Shortcuts in the table and tree views used to stop working the moment you clicked anything outside the grid. The cell still looked selected, but every key did nothing, including the shortcut printed on the button you had just clicked. That's fixed, and SQL tables have caught up with MongoDB.

What stopped happening

  • Shortcuts no longer go dead after a toolbar click. Click a toolbar button, the pager or an empty spot, and your shortcuts keep working on the table you're looking at. This covers MongoDB and SQL, in both table and tree view.
  • One key press does one thing. Some shortcuts, including Refresh, used to run twice on a single press.
  • Menus and tooltips show your own keys. If you changed a shortcut, the right-click menus and toolbar tooltips now show the key you chose, not the original one.

What you can do

  • Edit SQL tables from the keyboard. The SQL row menu showed these shortcuts for a long time, but the keys did nothing. Now they work, with the same keys as MongoDB:
  • Insert row: ⌘/Ctrl+I
  • Edit row: ⌘/Ctrl+Shift+E
  • Delete rows: Shift+Delete
  • Batch update: ⌘/Ctrl+U
  • Batch delete: ⌘/Ctrl+D
  • Refresh: ⌘/Ctrl+R
  • Expand / collapse columns: Alt+β†’ and Alt+←
  • Run a SQL table's query with your own Run key. It follows the same setting as MongoDB's Run, so changing it once changes it in both.

In the Keyboard Shortcuts window

  • One list for MongoDB and SQL. Most actions use the same key in both, so they now share a row. Only the exceptions are labelled, such as SQL only or MongoDB only.
  • Moving around grids is listed too: arrows, Shift+arrows to extend a selection, Page Up/Down, Home/End, Tab, and Esc.
  • See what has no shortcut. A No shortcut filter lists every command without a key, and those rows are highlighted so you can assign one.
  • Know why a key isn't free, right where you're editing. While you set a key, the row tells you:
  • if it's taken by something that can't be changed (and stops you)
  • if it's used by another command, and whether that command keeps another shortcut or ends up with none
  • if it means something else elsewhere in the app, and which one wins
  • if it's already this command's key

Compare & Sync: Dry Run

Before, you had to trust a sync to do what you expected. Now you can see the plan first, for SQL and MongoDB alike.

What you can see

  • "Dry run β€” nothing has been written." Every sync can be checked first: a progress bar with X of Y checked, then the counts of inserts, updates and deletes, the first 500 changes, and the rows that will be skipped and why (already in sync, changed since the compare, gone from the source).
  • The exact statements. For SQL, the real INSERT / UPDATE / DELETE statements in your database's own syntax. For MongoDB, the equivalent shell commands, with types like NumberDecimal, ISODate and ObjectId kept intact.
  • A warning when your database commits row by row, so you know a sync on it can't be rolled back halfway.

What you can do

  • Click Sync plan and the dry run starts right away. What it shows is exactly what Start sync will do.
  • Stop a dry run that's taking too long.
  • Download the full plan as .sql or .js to review, share or run yourself. Large plans work too; the file is ready for 30 minutes.
  • Sync progress no longer gets stuck at 0 when you have several VisuaLeaf tabs open in a browser. The dialog finishes on its own.

Charts

Making a chart used to start with an empty canvas and a list of field boxes. Now it starts with a chart.

What you can see

  • A chart already drawn. A new chart from a query, pipeline or SQL picks fields and a chart type that fit your results: a line over dates, a bar or pie over a repeating category, a gauge for a single value, or a table otherwise. Saved charts are left as they were.
  • Boxes named after the chart on screen. X axis / Y axis for bar, line and area; Slices / Size for pie; X axis / Cells / Y axis for heatmaps; Value for gauges.
  • "Goes here" / "Will change." Hovering a field highlights the box it would fill, so nothing changes by surprise.
  • Clearer labels: "Count of rows" instead of the column used to count, a footer that says what the chart shows, and gauges with negative values get a scale that reaches below zero.

What you can do

  • Click a box to pick it, then click fields to try them in that box one after another. Done, Esc or a click elsewhere lets go.
  • Clicking a field never replaces a box you already set. It fills an empty one, and never one the chart type doesn't use, so a chart no longer switches type on its own.
  • Use the "+ …" in an empty box to pick from that box's own field list.

Fixed: switching a chart from Table back to Bar no longer leaves it blank.


Grids & Query Builder

What you can do

  • Select several cells. Cmd/Ctrl+click adds a cell, Cmd/Ctrl+drag adds a block. Copy keeps each value in its place, and Cmd/Ctrl+C works after selecting rows by their row numbers.
  • Drag a selection into the query builder and every selected cell becomes a condition.
  • MongoDB: several values of one field become a single $in, duplicates listed once, including sub-documents and arrays. Dropping several values on an existing condition turns it into an $in.
  • SQL: empty cells become IS NULL (or IN … OR IS NULL alongside other values). Dates no longer shift back a day, timestamps keep their fractions of a second, and PostgreSQL arrays, JSON, XML and money columns filter correctly.
  • Set as Query (MongoDB table view): the filters in Columns & Filters used to hide rows on the loaded page only. Now they can run against the whole collection, added to your current query, with numbers, dates, ObjectIds and booleans kept as their real types.

What stopped happening

  • Dragging a value from the MongoDB table into the query builder no longer stutters, even after switching windows.
  • A SQL grid no longer goes blank after a re-run or refresh while scrolled down. A one-row result now shows its row.
  • Disabled conditions, sorts and projections stay put in the MongoDB query builder when you switch tabs.

Improvements

Oracle

  • Table Management opens a table in about 0.2 seconds instead of 4–9. Measured on every table in a 56-table schema, with identical results before and after.
  • Foreign keys come from the right table. If another schema had a table with the same name, its foreign keys could show up in yours. They no longer do.

Database Menu

  • Sections for MongoDB, SQL and Redis. The Database menu used to hold MongoDB tools only. It now also has:
  • SQL: Tables Info, Query Profiling, System Monitor, Users & Roles, Server Info
  • Redis: Key Browser, Server Info, Users & Security

These screens already existed, but you could only find them through the command palette or the sidebar. The native macOS menu has the same sections.

ER Diagrams

  • Hide one relationship's name: click a line and choose Hide name. The line and its 1 / N markers stay, and hovering shows the name. A toolbar button hides or shows all names at once, and hidden names stay hidden in PNG and SVG exports.
  • N-1 relationships behave: picking N-1 highlights the button, and switching back to 1-N works.
  • Each database shows its own logo in the Blank Schema menu.
  • Relationship lines stay drawn when the AI assistant arranges a diagram.

Connections

  • One login attempt per Test. A wrong password on Test Connection used to try several times (around nine on Oracle, enough to lock an account) and could also break another open connection to the same database. Now it's one attempt.
  • Pasting a SQL connection URL keeps its TLS settings. Options like sslmode=verify-full used to be dropped silently. They now land in Advanced driver properties, and anything that can't be kept is listed instead of vanishing.
  • The real reason, not a puzzle. ClickHouse says why a connection failed (wrong password, refused, unknown host) and gives up on a firewalled host in under a minute instead of five. A wrong MySQL 8 password says "Access denied". Oracle, SQL Server, Azure SQL, HSQLDB, QuestDB, RisingWave and CockroachDB get the right hint for timeouts, unknown hosts and bad logins.

AI Assistant

  • Charts the assistant creates show data straight away and carry a title. Saving one updates it instead of making a copy.
  • Ask the assistant to build a dashboard. It can list your dashboards, create one laid out from your saved charts, add a chart to an existing one, open a dashboard so you see it drawn, and delete one (after you confirm). Deleting a dashboard keeps its charts.
  • Ask it to put a chart on your home page, too.

Data Transfer

  • The task list no longer keeps the CPU busy. With a long task history, VisuaLeaf used over 160% CPU just showing the list. It now uses about 6% of one core.
  • Import Data from File(s)… is on a schema's right-click menu.

Plans & Trial

  • A clear heads-up before your Pro trial ends. Three days before, VisuaLeaf shows the days left and compares Community, Basic and Pro, so you can see which features you'd keep.
  • Chart and dashboard tabs show a Pro page when your plan doesn't include them, and unlock themselves when it does.
  • Trials no longer show as expired by mistake on a computer whose clock was once set ahead. The licence server's answer now wins while you're online.

Help

  • Help β†’ Video Tutorials opens visualeaf.com/videos.

Notes

  • Free plan: saving is now limited to 5 schemas and 3 ER diagrams. Anything you already have stays, and editing, renaming or moving existing ones is never blocked. Only creating a new one over the limit asks you to upgrade.
  • Home page: nothing changes until you click Customize. The default page looks like the home page you already know.
  • If you changed any table shortcuts before this release, check them in Keyboard Shortcuts. SQL tables now have their own editing keys, and the window tells you about any overlap with your own choices.
Read the full v1.0.2714 release notes

v1.0.2686

November 3, 2026

VisuaLeaf v1.0.2686 Release Notes

Release Date: November 3, 2026

Covers 1.0.2678 – 1.0.2686.


Headline: Safer Batch Operations, Connection Errors That Say What's Wrong, and Smoother Linux Updates

Batch operations now only touch the rows you picked. In MongoDB, choosing "Selected Documents" with
nothing selected could update or delete the whole collection. That can no longer happen. **SQL Batch
Update works for the first time**, on every engine.

Failed connections now tell you why. Instead of "500 Server Error" or a 30-second wait, you get the
server's real reason and a short What to check hint. A MongoDB connection to a server that's down
fails in seconds. When a MySQL or TiDB server requires encryption, VisuaLeaf turns it on and tells you.

Also shipping: insert a MongoDB document from a template built from the collection's validation
rules, validation rules shown on every field, accurate PostgreSQL row counts in Table Stats, and
Linux updates that close and reopen the app for you.


Batch Operations

What stopped happening

  • No more whole-collection updates or deletes by accident. In MongoDB's grid and tree views, "Selected
  • Documents" is greyed out when nothing is selected, and the dialog opens on All documents or
    Current search instead. Before, an empty selection meant every document in the collection.

  • Selecting a field in the tree view targets only its own document. It used to also hit every other
  • document with the same value in that field.

  • SQL Batch Update works. It used to fail every time. Values with apostrophes, booleans and dates now
  • save correctly on every engine.

  • SQL "All rows" is honest about what it does. With a filter on, it now says it will empty the whole
  • table instead of showing the filtered row count. The filter preview shows just the condition.

  • A filter can never quietly widen to the whole table. Batch update and delete refuse a request that
  • names no rows, no filter and no "all rows" choice.

  • The grid tidies up after a delete. The Batch Delete dialog closes when it succeeds, the deleted rows
  • are no longer highlighted, and a document whose _id is 0 is no longer skipped.

  • CrateDB grids stop showing rows you just deleted.

What you can see

  • Redesigned MongoDB Batch Insert, Update and Delete dialogs that match the rest of the app. Their
  • options stay on one line in narrow windows.


MongoDB Validation Rules

What you can see

  • Each field's rules in the Schema Builder. View mode shows rules as small chips (β‰₯ 2000,
  • len 1–120, items β‰₯ 1, unique, the pattern, and so on), with a tooltip explaining each one.
    (SOZO-1693)

  • Rules in the data viewer. Opening a value shows a Rules chip listing that field's rules. Opening a
  • whole document lists every field that has rules, by name. This only appears for collections with a real
    validator, never for a schema guessed from sample documents.

  • A heads-up for "constraint" validation. The JSON Schema Builder explains when a collection's rules
  • can't be changed because of its validation level. (SOZO-1695)

What you can do

  • Insert a document from a template. Batch Insert has a new Template option that builds a starting
  • document that passes the collection's validator: required fields first (also listed above the editor),
    allowed values, numbers within range, and strings and arrays long enough. Without a validator, the
    template comes from fields found in existing documents. (SOZO-1694)

  • Template opens by default on a collection with a validator, since an empty document would almost
  • always be rejected. You can still switch to Custom JSON.


Connections

What you can see

  • The real reason a connection failed, plus a What to check hint, on every MongoDB, SQL and Redis
  • error. The SQL connection form used to show "500 Server Error" for every failed test.

  • Clearer wording for common problems: a wrong SSH key passphrase, a jump host that refused the
  • connection, a connection timeout, an IPv6-only host, or a wrong replica set name. The connection form
    keeps these messages instead of replacing them with a generic one.

  • A stopped server is no longer called a TLS problem. When a MongoDB server simply isn't running (a
  • stopped Docker container, for example), the error says so.

  • "Connected with TLS." When a MySQL, MariaDB or TiDB server refuses unencrypted connections, VisuaLeaf
  • retries with encryption and tells you, in the test result and as a toast when you connect. The form saves that choice
    for you, and never overrides a TLS mode you picked yourself.

What you can do

  • Find out in seconds that a MongoDB server is down. A connection ends as soon as every server has failed, or a
  • mongodb+srv address can't be found, instead of waiting 30 seconds.

  • Connect to MySQL and TiDB servers that require encryption without changing any settings.
  • Use certificate files with PostgreSQL and MySQL in the desktop app. CA and client certificate
  • settings in Advanced driver properties are accepted when you save, instead of being refused.
    PostgreSQL's channel binding setting is accepted too.

  • Reach Neon through an SSH tunnel. It used to fail every time. **Full certificate checking also
  • works through a tunnel now** for PostgreSQL and MySQL; before, it always failed.


Improvements

Table Stats

  • PostgreSQL row counts you can trust. A table that had never been analyzed could show 0 rows forever
  • even when full of data. Those tables are now counted exactly. Very large ones show an estimate as ~N,
    with a hint to run ANALYZE for an exact number, in both the Tables Info grid and Table Management's
    Storage tab. (SOZO-1690)

Updates and Quitting

  • Linux updates close and reopen the app for you. Updating from inside the app on Linux (.deb or .rpm)
  • now asks first (Install and Reopen / Not Now, plus the usual question about unsaved work), closes
    VisuaLeaf while the installer runs, and starts the new version when it's done. Before, the app kept
    running while its own files were replaced, which could make it crash, or stop loading data until you restarted it.

  • Reopening right after quitting works. Quitting now waits for VisuaLeaf to fully shut down. Before,
  • reopening within a few seconds could fail to start, or take much longer while it cleaned up.

  • Clicking the app while it's closing reopens it once it has finished closing, instead of doing
  • nothing. This works when you launch it again on Windows and Linux, and from the Dock on macOS.


Notes

  • Linux users: the in-app update now closes VisuaLeaf while the installer runs. Save your work when it
  • asks; the new version opens on its own when the install finishes. If you cancel the installer,
    VisuaLeaf won't reopen by itself. Start it yourself. This starts with the update after this one. For
    this update, close VisuaLeaf before clicking Install in your system installer.

Read the full v1.0.2686 release notes

v1.0.2715-mongodb34

November 3, 2026

VisuaLeaf v1.0.2715 Release Notes

Release Date: November 6, 2026

Covers everything since v1.0.2686.


Headline: Make the Home Page Yours, See a Sync Before It Runs, and Shortcuts That Always Work

The biggest change in this release is the first thing you see: the home page is now yours to arrange. Add pages, drag and resize tiles, pick a preset, and put live dashboards, single charts and your favorite tables right where you open the app.

Also shipping:

  • A dry run for every Compare sync. See exactly what a sync would insert, update and delete, with live progress and a Stop button, and download the full plan as .sql or .js before anything is written.
  • Keyboard shortcuts that always work. Clicking a toolbar button no longer switches off your table shortcuts, SQL tables get the same editing keys as MongoDB, and the Keyboard Shortcuts window tells you which keys are free.
  • Charts that draw themselves. A new chart from a query opens with a sensible chart already on screen, and the field boxes are named after what they control (X axis, Y axis, Slices…).
  • Select several cells and drag them into the query builder. Each cell becomes the right filter for its type, in both SQL and MongoDB.
  • Oracle Table Management opens tables in about 0.2 seconds, down from 4–9 seconds.
  • Connection errors that don't lock you out. A wrong password on Test Connection is now one login attempt, not nine.
  • A leaner, more private AI assistant. Each message sends a fraction of what it used to, sample data is capped and secrets are blanked, and a new AI Request Log shows exactly what was sent.
  • Oracle fixes: JSON stored in CLOB columns now expands and edits like JSON, and following a foreign key no longer fails with ORA-00907.

Customizable Home Page

The home page used to be a fixed layout. Now you can shape it around how you work: a page for your daily dashboards, another for the tables you open every morning.

What you can see

  • Pages. Name your own tabs on the home page, each with its own tiles. The tab strip only shows up once you have a second page, and VisuaLeaf reopens the page you used last.
  • Live dashboards and charts on the home page. Any saved dashboard, or a single saved chart, can be a tile. A chart tile is drawn exactly as it looks on a dashboard and grows when you make the tile taller. Each tile loads only when you scroll to it, waits politely while its connections are closed (with a Connect & show button), and fills in as soon as they open.
  • Favorites. Right-click any collection or table and choose Add to Home Favorites to open it in one click from the home page.
  • Tiles for every tool, including Query Profiler, Schema Manager, Compare Data, Saved Items, Data Transfer, New Connection and Connection Manager, plus heading tiles you can label yourself.
  • A new default home page, Classic. Your shortcuts are grouped under headings (Work with your data, Pick up where you left off, Build and move data, Get going) with your recent tabs alongside. The previous layout is still one click away as the Bento preset.
  • Presets. Classic, Bento, Bento Sections, My work, Dashboard first, Calm, Original (the first VisuaLeaf home page), Sidebar and Focus, each previewed as a thumbnail before you apply it.
  • Three tile looks: Compact, Card and Preview, in ten accent colors. Tint all recolors a whole page at once. Tiles adapt to their size: short tiles become a single line, tall ones show the feature's picture.

What you can do

  • Click Customize, then drag, resize, remove and add tiles back. Your layout is saved automatically.
  • Undo and redo while you customize. ⌘Z / Ctrl+Z steps back through moves, resizes, removed tiles or pages and applied presets; β‡§βŒ˜Z / Ctrl+Shift+Z steps forward. A whole drag counts as one step.
  • Turn off Snap together on a page to place tiles freely, with gaps where you want them. Tidy up packs the page again.
  • Move several tiles at once. Shift/Ctrl-click or drag a box around tiles to select them, then drag one and the rest follow. The preview shows where every tile will land before you let go. Resizing one selected tile resizes them all.
  • Change color, look or remove a whole selection in one step.
  • Work in two windows. A change in one window shows up in the other right away, and an older window can no longer overwrite your newer layout.
  • Pick a chart for a Chart tile, then Refresh it in place or Open it as a tab. A closed connection, a deleted or unshared chart, or a plan without charts each gets a clear message instead of a blank tile.
  • Ask the AI assistant to do it. Through the MCP connection, an assistant can create pages, apply presets, and add dashboards, charts, tiles and favorites for you.

Keyboard Shortcuts

Shortcuts in the table and tree views used to stop working the moment you clicked anything outside the grid. The cell still looked selected, but every key did nothing, including the shortcut printed on the button you had just clicked. That's fixed, and SQL tables have caught up with MongoDB.

What stopped happening

  • Shortcuts no longer go dead after a toolbar click. Click a toolbar button, the pager or an empty spot, and your shortcuts keep working on the table you're looking at. This covers MongoDB and SQL, in both table and tree view.
  • One key press does one thing. Some shortcuts, including Refresh, used to run twice on a single press.
  • Menus and tooltips show your own keys. If you changed a shortcut, the right-click menus and toolbar tooltips now show the key you chose, not the original one.

What you can do

  • Edit SQL tables from the keyboard. The SQL row menu showed these shortcuts for a long time, but the keys did nothing. Now they work, with the same keys as MongoDB:
  • Insert row: ⌘/Ctrl+I
  • Edit row: ⌘/Ctrl+Shift+E
  • Delete rows: Shift+Delete
  • Batch update: ⌘/Ctrl+U
  • Batch delete: ⌘/Ctrl+D
  • Refresh: ⌘/Ctrl+R
  • Expand / collapse columns: Alt+β†’ and Alt+←
  • Run a SQL table's query with your own Run key. It follows the same setting as MongoDB's Run, so changing it once changes it in both.

In the Keyboard Shortcuts window

  • One list for MongoDB and SQL. Most actions use the same key in both, so they now share a row. Only the exceptions are labelled, such as SQL only or MongoDB only.
  • Moving around grids is listed too: arrows, Shift+arrows to extend a selection, Page Up/Down, Home/End, Tab, and Esc.
  • See what has no shortcut. A No shortcut filter lists every command without a key, and those rows are highlighted so you can assign one.
  • Know why a key isn't free, right where you're editing. While you set a key, the row tells you:
  • if it's taken by something that can't be changed (and stops you)
  • if it's used by another command, and whether that command keeps another shortcut or ends up with none
  • if it means something else elsewhere in the app, and which one wins
  • if it's already this command's key

Compare & Sync: Dry Run

Before, you had to trust a sync to do what you expected. Now you can see the plan first, for SQL and MongoDB alike.

What you can see

  • "Dry run β€” nothing has been written." Every sync can be checked first: a progress bar with X of Y checked, then the counts of inserts, updates and deletes, the first 500 changes, and the rows that will be skipped and why (already in sync, changed since the compare, gone from the source).
  • The exact statements. For SQL, the real INSERT / UPDATE / DELETE statements in your database's own syntax. For MongoDB, the equivalent shell commands, with types like NumberDecimal, ISODate and ObjectId kept intact.
  • A warning when your database commits row by row, so you know a sync on it can't be rolled back halfway.

What you can do

  • Click Sync plan and the dry run starts right away. What it shows is exactly what Start sync will do.
  • Stop a dry run that's taking too long.
  • Download the full plan as .sql or .js to review, share or run yourself. Large plans work too; the file is ready for 30 minutes.
  • Sync progress no longer gets stuck at 0 when you have several VisuaLeaf tabs open in a browser. The dialog finishes on its own.

Charts

Making a chart used to start with an empty canvas and a list of field boxes. Now it starts with a chart.

What you can see

  • A chart already drawn. A new chart from a query, pipeline or SQL picks fields and a chart type that fit your results: a line over dates, a bar or pie over a repeating category, a gauge for a single value, or a table otherwise. Saved charts are left as they were.
  • Boxes named after the chart on screen. X axis / Y axis for bar, line and area; Slices / Size for pie; X axis / Cells / Y axis for heatmaps; Value for gauges.
  • "Goes here" / "Will change." Hovering a field highlights the box it would fill, so nothing changes by surprise.
  • Clearer labels: "Count of rows" instead of the column used to count, a footer that says what the chart shows, and gauges with negative values get a scale that reaches below zero.

What you can do

  • Click a box to pick it, then click fields to try them in that box one after another. Done, Esc or a click elsewhere lets go.
  • Clicking a field never replaces a box you already set. It fills an empty one, and never one the chart type doesn't use, so a chart no longer switches type on its own.
  • Use the "+ …" in an empty box to pick from that box's own field list.

Fixed: switching a chart from Table back to Bar no longer leaves it blank.


Grids & Query Builder

What you can do

  • Select several cells. Cmd/Ctrl+click adds a cell, Cmd/Ctrl+drag adds a block. Copy keeps each value in its place, and Cmd/Ctrl+C works after selecting rows by their row numbers.
  • Drag a selection into the query builder and every selected cell becomes a condition.
  • MongoDB: several values of one field become a single $in, duplicates listed once, including sub-documents and arrays. Dropping several values on an existing condition turns it into an $in.
  • SQL: empty cells become IS NULL (or IN … OR IS NULL alongside other values). Dates no longer shift back a day, timestamps keep their fractions of a second, and PostgreSQL arrays, JSON, XML and money columns filter correctly.
  • Set as Query (MongoDB table view): the filters in Columns & Filters used to hide rows on the loaded page only. Now they can run against the whole collection, added to your current query, with numbers, dates, ObjectIds and booleans kept as their real types.

What stopped happening

  • Deleting rows with a column filter on deletes the rows you selected. In a filtered SQL table, rows picked by their row numbers and deleted from the menu or with Shift+Delete could remove different rows. It now deletes exactly what you selected.
  • Dragging a value from the MongoDB table into the query builder no longer stutters, even after switching windows.
  • A SQL grid no longer goes blank after a re-run or refresh while scrolled down. A one-row result now shows its row.
  • Disabled conditions, sorts and projections stay put in the MongoDB query builder when you switch tabs.

Improvements

Oracle

  • JSON in CLOB columns expands like JSON. A CLOB or VARCHAR2 column that Oracle declares as JSON (IS JSON) now opens into its fields in the table and tree views, and you can edit a single field in place. Text columns that aren't declared as JSON stay text, even when their content looks like JSON. (SOZO-1717)
  • Following a foreign key works. Clicking the eye icon on a foreign key used to fail with "ORA-00907: missing right parenthesis". It now opens the related row, and the foreign-key check in the row and batch editors works on Oracle too. (SOZO-1718)
  • Table Management opens a table in about 0.2 seconds instead of 4–9. Measured on every table in a 56-table schema, with identical results before and after.
  • Foreign keys come from the right table. If another schema had a table with the same name, its foreign keys could show up in yours. They no longer do.

Database Menu

  • Sections for MongoDB, SQL and Redis. The Database menu used to hold MongoDB tools only. It now also has:
  • SQL: Tables Info, Query Profiling, System Monitor, Users & Roles, Server Info
  • Redis: Key Browser, Server Info, Users & Security

These screens already existed, but you could only find them through the command palette or the sidebar. The native macOS menu has the same sections.

ER Diagrams

  • Hide one relationship's name: click a line and choose Hide name. The line and its 1 / N markers stay, and hovering shows the name. A toolbar button hides or shows all names at once, and hidden names stay hidden in PNG and SVG exports.
  • N-1 relationships behave: picking N-1 highlights the button, and switching back to 1-N works.
  • Each database shows its own logo in the Blank Schema menu.
  • Relationship lines stay drawn when the AI assistant arranges a diagram.

Connections

  • One login attempt per Test. A wrong password on Test Connection used to try several times (around nine on Oracle, enough to lock an account) and could also break another open connection to the same database. Now it's one attempt.
  • Pasted MySQL and PostgreSQL URLs with common TLS options connect. useSSL=true, sslMode=PREFERRED and Heroku-style sslfactory=…NonValidatingFactory are translated into settings the connection understands, instead of making the connection fail.
  • Pasting a SQL connection URL keeps its TLS settings. Options like sslmode=verify-full used to be dropped silently. They now land in Advanced driver properties, and anything that can't be kept is listed instead of vanishing.
  • The real reason, not a puzzle. ClickHouse says why a connection failed (wrong password, refused, unknown host) and gives up on a firewalled host in under a minute instead of five. A wrong MySQL 8 password says "Access denied". Oracle, SQL Server, Azure SQL, HSQLDB, QuestDB, RisingWave and CockroachDB get the right hint for timeouts, unknown hosts and bad logins.

AI Assistant

  • Much smaller requests. The assistant now sends a compact outline of your collection's fields instead of the full schema. On a 100-field collection that is about 2,500 characters instead of 62,000, on every message.
  • Sample data is capped and scrubbed. When "Send sample data" is on, at most 5 documents go out, long values and arrays are trimmed, and fields whose names suggest a secret (password, token, API key, SSN…) are replaced with [redacted].
  • See exactly what was sent. Settings β†’ AI β†’ AI Request Log lists every request with the model, the full text sent and received, and the token count. It is kept in memory only and cleared when VisuaLeaf restarts.
  • Charts the assistant creates show data straight away and carry a title. Saving one updates it instead of making a copy.
  • Ask the assistant to build a dashboard. It can list your dashboards, create one laid out from your saved charts, add a chart to an existing one, open a dashboard so you see it drawn, and delete one (after you confirm). Deleting a dashboard keeps its charts.
  • Ask it to put a chart on your home page, too.

Schema Editor

  • Saving a new JSON schema twice no longer creates a copy on a VisuaLeaf server: later saves update the schema you are editing.

Data Transfer

  • The task list no longer keeps the CPU busy. With a long task history, VisuaLeaf used over 160% CPU just showing the list. It now uses about 6% of one core.
  • Import Data from File(s)… is on a schema's right-click menu.

Plans & Trial

  • A clear heads-up before your Pro trial ends. Three days before, VisuaLeaf shows the days left and compares Community, Basic and Pro, so you can see which features you'd keep.
  • Chart and dashboard tabs show a Pro page when your plan doesn't include them, and unlock themselves when it does.
  • Trials no longer show as expired by mistake on a computer whose clock was once set ahead. The licence server's answer now wins while you're online.

Help

  • Help β†’ Video Tutorials opens visualeaf.com/videos.

Notes

  • Free plan: saving is now limited to 5 schemas and 3 ER diagrams. Anything you already have stays, and editing, renaming or moving existing ones is never blocked. Only creating a new one over the limit asks you to upgrade.
  • Home page: if you never customized it, your home page now opens in the new Classic layout. The previous layout is the Bento preset under Customize. Home pages you already customized are unchanged.
  • If you changed any table shortcuts before this release, check them in Keyboard Shortcuts. SQL tables now have their own editing keys, and the window tells you about any overlap with your own choices.
Read the full v1.0.2715-mongodb34 release notes

v1.0.2677

November 2, 2026

VisuaLeaf v1.0.2677 Release Notes

Release Date: November 2, 2026

Covers 1.0.2640 – 1.0.2677.


Headline: BI-Style Dashboards and Charts, Parquet Export, and Copies That Keep Every Type

Dashboards now work like a BI tool. Set filters once for the whole dashboard, click a bar or slice to
filter every other chart, and build charts by dropping fields into Group by / Measure / Color / Filters
boxes, with totals that are exact over every document. Every chart has a new look that works in light
and dark mode, and the dashboard editor has a cleaner design with edges you can actually grab to resize.

Parquet is a new export format for every SQL engine and MongoDB. **Data Transfer now keeps column types
and values exact** across all 29 SQL engines, MongoDB and files. A copy between two PostgreSQL or two MySQL
databases keeps each column's own type, and starting a task takes a fraction of a second even with a long
task history.

Also shipping: a redesigned Connection Manager that reads like a folder tree, **SSH tunnels that check
the server's identity, a redesigned Replica Set page built around health at a glance, Drop Schema**
for SQL, long SQL scripts that finish instead of freezing the editor, a two-row SQL Editor toolbar,
drag jobs between tasks in the Task Manager, and fixes to the ERD, popouts and licence activation.


Dashboards and Charts

What you can see

  • One modern look for every chart. Charts use VisuaLeaf's own palette in both light and dark mode, so a
  • dashboard reads as one set. Pies and donuts have gaps between slices and show the total in the middle of
    a donut. Heatmaps and calendars use a single colour scale instead of a rainbow. Gauges, maps, radar and
    candlestick charts no longer show bright bands or dark slabs in the wrong mode. Charts already on a
    dashboard pick up the new look when it opens. Colours you picked yourself are kept.

  • A redesigned dashboard editor and viewer. You edit the title and description directly in place, and
  • active filters show as chips under the header. Cards are cleaner, sit on a quiet dot-grid canvas, and
    show edit and delete only on hover. An empty dashboard invites you to start.

  • Pivot charts show a real pivot grid, with row values down the side, column values across the top,
  • and totals. Tree charts show parents with their children indented underneath.

  • Radar charts scale to your data instead of always running 0–100, so ratings, prices and counts no
  • longer shrink to a dot or run off the edge.

  • Charts fit narrow cards. In a narrow card, a pie moves its legend underneath, long radar names wrap,
  • and funnel and heatmap labels no longer overlap.

  • A clear row-limit warning. It tells you when totals use only the first 1,000 rows and offers to use
  • all rows.

What you can do

  • Filter the whole dashboard. Define filters once (Edit β†’ Filters), choose which charts each one
  • applies to, and pick values loaded from your data.

  • Click to filter. Click a bar or slice and the other charts filter to match. The selection survives
  • tab switches, and each chart shows a badge when a filter applies to it.

  • Build charts from fields. Drop fields into Group by, Measure, Color and Filters. Show me suggests
  • chart types, and More types reaches every chart type, telling you what each one needs. Gauge and
    heatmap now work from the boxes too.

  • Bring older charts along. Saved query charts open in the boxes when every part maps exactly. A banner
  • shows what moved, with Keep my query instead. Charts that keep their own query (pipelines, SQL) get
    the same boxes over their result columns.

  • SQL charts on dashboards run their query live, so dashboard filters and parameters reach them.
  • Resize cards from any edge. Resize handles now sit on the card's edges and bottom corners, with
  • larger grab areas.

Fixes

  • A dashboard filter narrows a chart instead of overriding it. A "delivered orders" chart stays on
  • delivered orders when you filter or click "pending" elsewhere on the dashboard, and two filters on the
    same field both apply.

  • Table, pivot and tree charts on a dashboard no longer show "No rows".
  • Queries typed in shell style ({ status: "delivered" }) are applied instead of silently ignored.
  • Swapping a widget's chart loads the new chart. Removing a widget respects Cancel, and new widgets land
  • below the last row instead of at the top-left.

  • Text such as "CUST-105050" is no longer mistaken for a date.
  • The dashboard's loading and export overlays cover only that dashboard, not the whole window.
  • "Keep my query instead" stays available after you switch tabs or reopen the app, and a converted
  • chart is only kept once you accept it. A duplicated chart doesn't offer to restore the original.

  • A converted chart shows the same totals it showed before, and converted bar, line and area charts
  • keep their categories Aβ†’Z, so a line over months or years no longer zigzags. Pie, donut, treemap and
    funnel stay largest first.


Connection Manager, Redesigned

What you can see

  • Three panes. Your views and projects on the left, your connections in the middle, and the details of
  • whatever you select on the right: host, port, user, database and when you last used it, with a big
    Connect button.

  • A folder tree. Projects, their environments and their connections nest like folders, with guide lines
  • showing what's inside what. Collapse a project or environment to tuck it away, and it stays collapsed.
    Every environment shows, so an empty project no longer opens to nothing.

  • Calmer rows. Each connection shows its engine, host and status (Connected, or when you last used it).
  • Connect and a β‹― menu appear when you hover, along with a drag handle so it's clear rows can be
    dragged.

  • Tidier connection forms. A centred column, name and database type side by side, compact text boxes
  • that stand out, and one main button: Save and connect. Test Connection and Export to URL are back in
    the top-right corner. A required field left empty stays red until you fill it in.

  • A cleaner Test Connection dialog. It shows the connection's name and how far the test has got, finished
  • steps get a green check, and the result is a single card: success, or the actual error.

  • A fresh look in both light and dark mode that matches the rest of the app. Menus stay inside the window on
  • small screens, and only one is open at a time.

What you can do

  • Select like a file browser. Click to select one connection, Ctrl/⌘-click to add or remove one, and
  • Shift-click to select a range.

  • Move connections where they belong. Drag them onto an environment, in the list or the sidebar, or use
  • Move to… in a connection's menu. While you drag, every environment you can drop on is outlined and the
    one under the pointer lights up. A message confirms each move.

  • Name things as you create them. New project and New environment ask for a name and a colour first, so
  • the list no longer fills up with "New Project"s. The new item is scrolled to and highlighted.

  • Manage projects and environments from the list. Right-click a project or environment, or use the
  • buttons that appear on hover: + adds an environment to a project or a connection to an environment,
    and Rename edits the name in place.

  • See why a connection failed. A failed connection now tells you why (a wrong password, an unreachable
  • host, a changed SSH server key) instead of just "Failed to connect".


SSH Tunnels Check the Server's Identity

  • VisuaLeaf now checks who it's talking to. The first time you connect through an SSH server, VisuaLeaf
  • remembers its host key, the way ssh does. If a later connection presents a different key, it's refused
    before your password or key is sent, so an impostor can't collect them. This applies to MongoDB, SQL and
    Redis connections over SSH.

  • Existing connections keep working. Their next connection trusts the server without asking.
  • The SSH tab shows the trusted key (its fingerprint) and has Forget trusted key for when a server
  • was rebuilt or its keys were replaced. If a key is refused, the error says so and points you there. In
    server mode, only an administrator can forget a key.


Export to Parquet

  • Parquet is a new export format next to CSV, XML and Excel, for a table, view, query, whole
  • database, collection or aggregation, from every SQL engine and MongoDB.

  • Columns keep their types. SQL columns keep their declared type, and a MongoDB field whose type changes
  • between documents is written as text instead of failing.

  • Parquet, XML and Excel tasks now show their file path in the task list instead of "Unknown".

Data Transfer: Every Type, Every Value

We ran every transfer, import and export through the app on 29 SQL engines and MongoDB, and compared every
target cell by cell. Here's what changed.

Copies between databases

  • Same-family copies keep the source's types. PostgreSQL β†’ PostgreSQL keeps arrays, intervals and
  • smallint. MySQL β†’ MySQL keeps its small integer, TEXT and BINARY types. SQL Server keeps money,
    nchar and uniqueidentifier.

  • Numbers stay exact. Oracle numbers without a set precision are no longer rounded to 4 decimals.
  • Floats are no longer widened (3.14159 stays 3.14159), and huge integers are no longer lost.

  • Dates and times stay put. MySQL keeps fractional seconds, and SQLite and DuckDB times are no
  • longer shifted by the server's timezone.

  • Copies that used to fail now finish: ClickHouse's mixed-type tables, PostgreSQL network types into
  • CockroachDB, DuckDB JSON tables, and engines that lack a CHAR or TIME type.

  • MongoDB β†’ SQL looks at 1,000 documents, not just the first, to pick each column's type. Fields that
  • appear later are kept, 12.5 is no longer stored as 12, and UUIDs no longer fail the task.

  • Dates copied from MongoDB into ClickHouse, and dates stored as text into SQLite, arrive intact.
  • A value is left empty only when it really can't be stored in its column, and the task says so.

Dumps and restores

  • A dump restores exactly on all 29 SQL engines, native or portable, as one file or one file per
  • table. Every type comes back, including enums, JSON, arrays, intervals, structs and zoned timestamps.

  • Restored tables keep counting from the right id, so the next insert doesn't collide with imported
  • rows.

  • A one-file-per-table dump only tidies up its own files. Re-running it removes files for tables
  • that no longer exist, but other dumps saved in the same folder are never touched.

  • A failed import file's log lists its first errors.

Files

  • CSV no longer shifts SQL dates by the server's timezone.
  • Arrays, long text and binary columns from H2 and HSQLDB are written as data, not as placeholder text.
  • MongoDB β†’ CSV, XML and Excel keeps fields missing from the first document, and Excel dates are no longer
  • shifted.

  • Imports into SQL Server keep Unicode (βœ“ no longer becomes ?).

Faster

  • Tasks start in a fraction of a second, even with a long task history. They used to take up to 30–50
  • seconds.


Replica Set: Health at a Glance

What you can see

  • A health strip across the top: members up, how many it can lose, oplog window and last election.
  • Member cards with colour-coded roles and how far each copy is behind the primary.
  • The tool cards are grouped under Health checks, Reads & writes and Activity tabs, with counts
  • when there's something to look at.

What you can do

  • Click a member to open its details in a side panel. With changes enabled, its actions are spelled out
  • in plain words: step down, make primary, maintenance, reconfigure, remove.

  • Enable changes is now the one prominent switch. When it's on, an amber bar holds the set-wide actions:
  • add member, rolling compact and force reconfig.


Improvements

SQL

  • Drop Schema from the schema's right-click menu (SOZO-1679). You type the name to confirm, see the
  • statement first, and can choose to drop everything inside it. On engines that can't do that, the dialog
    says the schema must be empty. Open tabs from that schema close afterwards.

  • Long scripts finish. A 40,000-statement dump that used to freeze the editor and never finish now runs
  • in about 15 seconds against a local database, with steady progress. Long runs of inserts are summarised
    as one line ("Statements 1–4,000: 4,000 rows affected"), and History isn't flooded with them.

  • A cleaner SQL Editor toolbar. It has two rows like the shell: the connection path on top, and
  • Run, Run Selected, Run to Line, the commit controls and Save Script above the editor. Neither row wraps,
    and in a narrow pane the buttons shrink to icons.

Task Manager

  • Drag a job onto another task in the sidebar to move it there.
  • Dragging tasks into folders works again, with no pause when the drag starts. A label follows the
  • pointer and Esc cancels.

  • The Cards layout is gone. Everyone uses List + details.
  • Success messages are solid green so you notice them.

ERD / Visual Schema

  • Refresh from Database no longer removes tables you've just drawn. A table that was never created is
  • marked as a draft, with a link to create it. Only tables actually dropped from the database are offered
    for removal, unticked by default. The Apply button says what it will do.

  • Tables removed with a card's βœ• no longer come back after reopening the diagram (SOZO-1684).
  • Required columns are marked with \* after their type, instead of nullable ones with a ?. Exported
  • PNG and SVG match.

  • Visual Schema from a Table Stats page opens that schema's diagram (SOZO-1691), with only that schema's
  • tables in the sidebar.

  • The column editors' key, nullable and foreign-key toggles say what they're set to and what a click does.

AI assistants (MCP)

  • AI assistants can create charts on SQL connections, and diagram edits keep primary keys however
  • they're given (SOZO-1683, SOZO-1685). A chart only reads: an assistant can't save one that would change
    data.

  • Settings β€Ί MCP Server hides the token in the client configuration until you reveal it. Copying still
  • copies the real token (SOZO-1686).

Everywhere else

  • Scrolling no longer stutters when the workspace autosaves. It waits until you pause.
  • A tab dragged from a popout back into the main window lands where you drop it, including on a panel
  • edge to split it.

  • Connection Manager: Select All ticks every row and switches to Select None.
  • Licence activation: if a licence can't be installed, the app retries, then keeps running on the
  • Community plan. The banner says why and lets you try again.


Notes

  • If an SSH server you connect through is rebuilt or its keys change, connections through it are refused
  • until you open the connection's SSH tab and choose Forget trusted key, then connect again.

  • If your SSH address leads to several servers with different keys (for example a bastion behind a load
  • balancer), connections may be refused some of the time. Let us know if this affects you.

Read the full v1.0.2677 release notes

v1.0.2706-mongodb34

November 2, 2026

VisuaLeaf v1.0.2706 Release Notes

Release Date: November 5, 2026


Headline: Make the Home Page Yours, See a Sync Before It Runs, and Charts That Draw Themselves

The biggest change in this release is the first thing you see: the home page is now yours to arrange. Add pages, drag and resize tiles, pick a preset, and put live dashboards and your favorite tables right where you open the app.

Also shipping:

  • A dry run for every Compare sync. See exactly what a sync would insert, update and delete, with live progress and a Stop button, and download the full plan as .sql or .js before anything is written.
  • Charts that draw themselves. A new chart from a query opens with a sensible chart already on screen, and the field boxes are named after what they control (X axis, Y axis, Slices…).
  • Select several cells and drag them into the query builder. Each cell becomes the right filter for its type, in both SQL and MongoDB.
  • Connection errors that don't lock you out. A wrong password on Test Connection is now one login attempt, not nine.

Customizable Home Page

The home page used to be a fixed layout. Now you can shape it around how you work: a page for your daily dashboards, another for the tables you open every morning.

What you can see

  • Pages. Name your own tabs on the home page, each with its own tiles. The tab strip only shows up once you have a second page, and VisuaLeaf reopens the page you used last.
  • Live dashboards on the home page. Any saved dashboard can be a tile. It loads only when you scroll to it, waits politely while its connections are closed (with a Connect & show button), and fills in as soon as they open.
  • Favorites. Right-click any collection or table and choose Add to Home Favorites to open it in one click from the home page.
  • Tiles for every tool, including Query Profiler, Schema Manager, Compare Data, Saved Items, Data Transfer, New Connection and Connection Manager, plus heading tiles you can label yourself.
  • Presets. Classic, Compact, Original (the first VisuaLeaf home page), Sections, Sidebar, Focus, Dashboard first and Calm, each previewed as a thumbnail before you apply it.
  • Three tile looks: Compact, Card and Preview, in ten accent colors. Tint all recolors a whole page at once. Tiles adapt to their size: short tiles become a single line, tall ones show the feature's picture.

What you can do

  • Click Customize, then drag, resize, remove and add tiles back. Your layout is saved automatically.
  • Turn off Snap together on a page to place tiles freely, with gaps where you want them. Tidy up packs the page again.
  • Move several tiles at once. Shift/Ctrl-click or drag a box around tiles to select them, then drag one and the rest follow. The preview shows where every tile will land before you let go. Resizing one selected tile resizes them all.
  • Change color, look or remove a whole selection in one step.
  • Work in two windows. A change in one window shows up in the other right away, and an older window can no longer overwrite your newer layout.
  • Ask the AI assistant to do it. Through the MCP connection, an assistant can create pages, apply presets, and add dashboards, tiles and favorites for you.

Compare & Sync: Dry Run

Before, you had to trust a sync to do what you expected. Now you can see the plan first, for SQL and MongoDB alike.

What you can see

  • "Dry run β€” nothing has been written." Every sync can be checked first: a progress bar with X of Y checked, then the counts of inserts, updates and deletes, the first 500 changes, and the rows that will be skipped and why (already in sync, changed since the compare, gone from the source).
  • The exact statements. For SQL, the real INSERT / UPDATE / DELETE statements in your database's own syntax. For MongoDB, the equivalent shell commands, with types like NumberDecimal, ISODate and ObjectId kept intact.
  • A warning when your database commits row by row, so you know a sync on it can't be rolled back halfway.

What you can do

  • Click Sync plan and the dry run starts right away. What it shows is exactly what Start sync will do.
  • Stop a dry run that's taking too long.
  • Download the full plan as .sql or .js to review, share or run yourself. Large plans work too; the file is ready for 30 minutes.
  • Sync progress no longer gets stuck at 0 when you have several VisuaLeaf tabs open in a browser. The dialog finishes on its own.

Charts

Making a chart used to start with an empty canvas and a list of field boxes. Now it starts with a chart.

What you can see

  • A chart already drawn. A new chart from a query, pipeline or SQL picks fields and a chart type that fit your results: a line over dates, a bar or pie over a repeating category, a gauge for a single value, or a table otherwise. Saved charts are left as they were.
  • Boxes named after the chart on screen. X axis / Y axis for bar, line and area; Slices / Size for pie; X axis / Cells / Y axis for heatmaps; Value for gauges.
  • "Goes here" / "Will change." Hovering a field highlights the box it would fill, so nothing changes by surprise.
  • Clearer labels: "Count of rows" instead of the column used to count, a footer that says what the chart shows, and gauges with negative values get a scale that reaches below zero.

What you can do

  • Click a box to pick it, then click fields to try them in that box one after another. Done, Esc or a click elsewhere lets go.
  • Clicking a field never replaces a box you already set. It fills an empty one, and never one the chart type doesn't use, so a chart no longer switches type on its own.
  • Use the "+ …" in an empty box to pick from that box's own field list.

Fixed: switching a chart from Table back to Bar no longer leaves it blank.


Grids & Query Builder

What you can do

  • Select several cells. Cmd/Ctrl+click adds a cell, Cmd/Ctrl+drag adds a block. Copy keeps each value in its place, and Cmd/Ctrl+C works after selecting rows by their row numbers.
  • Drag a selection into the query builder and every selected cell becomes a condition.
  • MongoDB: several values of one field become a single $in, duplicates listed once, including sub-documents and arrays. Dropping several values on an existing condition turns it into an $in.
  • SQL: empty cells become IS NULL (or IN … OR IS NULL alongside other values). Dates no longer shift back a day, timestamps keep their fractions of a second, and PostgreSQL arrays, JSON, XML and money columns filter correctly.
  • Set as Query (MongoDB table view): the filters in Columns & Filters used to hide rows on the loaded page only. Now they can run against the whole collection, added to your current query, with numbers, dates, ObjectIds and booleans kept as their real types.

What stopped happening

  • Dragging a value from the MongoDB table into the query builder no longer stutters, even after switching windows.
  • A SQL grid no longer goes blank after a re-run or refresh while scrolled down. A one-row result now shows its row.
  • Disabled conditions, sorts and projections stay put in the MongoDB query builder when you switch tabs.

Improvements

ER Diagrams

  • Hide one relationship's name: click a line and choose Hide name. The line and its 1 / N markers stay, and hovering shows the name. A toolbar button hides or shows all names at once, and hidden names stay hidden in PNG and SVG exports.
  • N-1 relationships behave: picking N-1 highlights the button, and switching back to 1-N works.
  • Each database shows its own logo in the Blank Schema menu.
  • Relationship lines stay drawn when the AI assistant arranges a diagram.

Connections

  • One login attempt per Test. A wrong password on Test Connection used to try several times (around nine on Oracle, enough to lock an account) and could also break another open connection to the same database. Now it's one attempt.
  • Pasting a SQL connection URL keeps its TLS settings. Options like sslmode=verify-full used to be dropped silently. They now land in Advanced driver properties, and anything that can't be kept is listed instead of vanishing.
  • The real reason, not a puzzle. ClickHouse says why a connection failed (wrong password, refused, unknown host) and gives up on a firewalled host in under a minute instead of five. A wrong MySQL 8 password says "Access denied". Oracle, SQL Server, Azure SQL, HSQLDB, QuestDB, RisingWave and CockroachDB get the right hint for timeouts, unknown hosts and bad logins.

AI Assistant

  • Charts the assistant creates show data straight away and carry a title. Saving one updates it instead of making a copy.

Data Transfer

  • The task list no longer keeps the CPU busy. With a long task history, VisuaLeaf used over 160% CPU just showing the list. It now uses about 6% of one core.
  • Import Data from File(s)… is on a schema's right-click menu.

Plans & Trial

  • A clear heads-up before your Pro trial ends. Three days before, VisuaLeaf shows the days left and compares Community, Basic and Pro, so you can see which features you'd keep.
  • Chart and dashboard tabs show a Pro page when your plan doesn't include them, and unlock themselves when it does.
  • Trials no longer show as expired by mistake on a computer whose clock was once set ahead. The licence server's answer now wins while you're online.

Help

  • Help β†’ Video Tutorials opens visualeaf.com/videos.

Notes

  • Free plan: saving is now limited to 5 schemas and 3 ER diagrams. Anything you already have stays, and editing, renaming or moving existing ones is never blocked. Only creating a new one over the limit asks you to upgrade.
  • Home page: nothing changes until you click Customize. The default page looks like the home page you already know.
Read the full v1.0.2706-mongodb34 release notes

v1.0.2664

November 1, 2026

VisuaLeaf v1.0.2664 Release Notes

Release Date: October 31 2026

Covers 1.0.2640 – 1.0.2664.


Headline: BI-Style Dashboards and Charts, Parquet Export, and Copies That Keep Every Type

Dashboards now work like a BI tool. Set filters once for the whole dashboard, click a bar or slice to
filter every other chart, and build charts by dropping fields into Group by / Measure / Color / Filters
boxes, with totals that are exact over every document. Every chart has a new look that works in light
and dark mode, and the dashboard editor has a cleaner design with edges you can actually grab to resize.

Parquet is a new export format for every SQL engine and MongoDB. **Data Transfer now keeps column types
and values exact** across all 29 SQL engines, MongoDB and files. A copy between two PostgreSQL or two MySQL
databases keeps each column's own type, and starting a task takes a fraction of a second even with a long
task history.

Also shipping: a redesigned Replica Set page built around health at a glance, Drop Schema for SQL,
long SQL scripts that finish instead of freezing the editor, a two-row SQL Editor toolbar,
drag jobs between tasks in the Task Manager, and fixes to the ERD, popouts and licence activation.


Dashboards and Charts

What you can see

  • One modern look for every chart. Charts use VisuaLeaf's own palette in both light and dark mode, so a
  • dashboard reads as one set. Pies and donuts have gaps between slices and show the total in the middle of
    a donut. Heatmaps and calendars use a single colour scale instead of a rainbow. Gauges, maps, radar and
    candlestick charts no longer show bright bands or dark slabs in the wrong mode. Charts already on a
    dashboard pick up the new look when it opens. Colours you picked yourself are kept.

  • A redesigned dashboard editor and viewer. You edit the title and description directly in place, and
  • active filters show as chips under the header. Cards are cleaner, sit on a quiet dot-grid canvas, and
    show edit and delete only on hover. An empty dashboard invites you to start.

  • Pivot charts show a real pivot grid, with row values down the side, column values across the top,
  • and totals. Tree charts show parents with their children indented underneath.

  • Radar charts scale to your data instead of always running 0–100, so ratings, prices and counts no
  • longer shrink to a dot or run off the edge.

  • Charts fit narrow cards. In a narrow card, a pie moves its legend underneath, long radar names wrap,
  • and funnel and heatmap labels no longer overlap.

  • A clear row-limit warning. It tells you when totals use only the first 1,000 rows and offers to use
  • all rows.

What you can do

  • Filter the whole dashboard. Define filters once (Edit β†’ Filters), choose which charts each one
  • applies to, and pick values loaded from your data.

  • Click to filter. Click a bar or slice and the other charts filter to match. The selection survives
  • tab switches, and each chart shows a badge when a filter applies to it.

  • Build charts from fields. Drop fields into Group by, Measure, Color and Filters. Show me suggests
  • chart types, and More types reaches every chart type, telling you what each one needs. Gauge and
    heatmap now work from the boxes too.

  • Bring older charts along. Saved query charts open in the boxes when every part maps exactly. A banner
  • shows what moved, with Keep my query instead. Charts that keep their own query (pipelines, SQL) get
    the same boxes over their result columns.

  • SQL charts on dashboards run their query live, so dashboard filters and parameters reach them.
  • Resize cards from any edge. Resize handles now sit on the card's edges and bottom corners, with
  • larger grab areas.

Fixes

  • A dashboard filter narrows a chart instead of overriding it. A "delivered orders" chart stays on
  • delivered orders when you filter or click "pending" elsewhere on the dashboard, and two filters on the
    same field both apply.

  • Table, pivot and tree charts on a dashboard no longer show "No rows".
  • Queries typed in shell style ({ status: "delivered" }) are applied instead of silently ignored.
  • Swapping a widget's chart loads the new chart. Removing a widget respects Cancel, and new widgets land
  • below the last row instead of at the top-left.

  • Text such as "CUST-105050" is no longer mistaken for a date.
  • The dashboard's loading and export overlays cover only that dashboard, not the whole window.

Export to Parquet

  • Parquet is a new export format next to CSV, XML and Excel, for a table, view, query, whole
  • database, collection or aggregation, from every SQL engine and MongoDB.

  • Columns keep their types. SQL columns keep their declared type, and a MongoDB field whose type changes
  • between documents is written as text instead of failing.

  • Parquet, XML and Excel tasks now show their file path in the task list instead of "Unknown".

Data Transfer: Every Type, Every Value

We ran every transfer, import and export through the app on 29 SQL engines and MongoDB, and compared every
target cell by cell. Here's what changed.

Copies between databases

  • Same-family copies keep the source's types. PostgreSQL β†’ PostgreSQL keeps arrays, intervals and
  • smallint. MySQL β†’ MySQL keeps its small integer, TEXT and BINARY types. SQL Server keeps money,
    nchar and uniqueidentifier.

  • Numbers stay exact. Oracle numbers without a set precision are no longer rounded to 4 decimals.
  • Floats are no longer widened (3.14159 stays 3.14159), and huge integers are no longer lost.

  • Dates and times stay put. MySQL keeps fractional seconds, and SQLite and DuckDB times are no
  • longer shifted by the server's timezone.

  • Copies that used to fail now finish: ClickHouse's mixed-type tables, PostgreSQL network types into
  • CockroachDB, DuckDB JSON tables, and engines that lack a CHAR or TIME type.

  • MongoDB β†’ SQL looks at 1,000 documents, not just the first, to pick each column's type. Fields that
  • appear later are kept, 12.5 is no longer stored as 12, and UUIDs no longer fail the task.

  • Dates copied from MongoDB into ClickHouse, and dates stored as text into SQLite, arrive intact.
  • A value is left empty only when it really can't be stored in its column, and the task says so.

Dumps and restores

  • A dump restores exactly on all 29 SQL engines, native or portable, as one file or one file per
  • table. Every type comes back, including enums, JSON, arrays, intervals, structs and zoned timestamps.

  • Restored tables keep counting from the right id, so the next insert doesn't collide with imported
  • rows.

  • A one-file-per-table dump only tidies up its own files. Re-running it removes files for tables
  • that no longer exist, but other dumps saved in the same folder are never touched.

  • A failed import file's log lists its first errors.

Files

  • CSV no longer shifts SQL dates by the server's timezone.
  • Arrays, long text and binary columns from H2 and HSQLDB are written as data, not as placeholder text.
  • MongoDB β†’ CSV, XML and Excel keeps fields missing from the first document, and Excel dates are no longer
  • shifted.

  • Imports into SQL Server keep Unicode (βœ“ no longer becomes ?).

Faster

  • Tasks start in a fraction of a second, even with a long task history. They used to take up to 30–50
  • seconds.


Replica Set: Health at a Glance

What you can see

  • A health strip across the top: members up, how many it can lose, oplog window and last election.
  • Member cards with colour-coded roles and how far each copy is behind the primary.
  • The tool cards are grouped under Health checks, Reads & writes and Activity tabs, with counts
  • when there's something to look at.

What you can do

  • Click a member to open its details in a side panel. With changes enabled, its actions are spelled out
  • in plain words: step down, make primary, maintenance, reconfigure, remove.

  • Enable changes is now the one prominent switch. When it's on, an amber bar holds the set-wide actions:
  • add member, rolling compact and force reconfig.


Improvements

SQL

  • Drop Schema from the schema's right-click menu (SOZO-1679). You type the name to confirm, see the
  • statement first, and can choose to drop everything inside it. On engines that can't do that, the dialog
    says the schema must be empty. Open tabs from that schema close afterwards.

  • Long scripts finish. A 40,000-statement dump that used to freeze the editor and never finish now runs
  • in about 15 seconds against a local database, with steady progress. Long runs of inserts are summarised
    as one line ("Statements 1–4,000: 4,000 rows affected"), and History isn't flooded with them.

  • A cleaner SQL Editor toolbar. It has two rows like the shell: the connection path on top, and
  • Run, Run Selected, Run to Line, the commit controls and Save Script above the editor. Neither row wraps,
    and in a narrow pane the buttons shrink to icons.

Task Manager

  • Drag a job onto another task in the sidebar to move it there.
  • Dragging tasks into folders works again, with no pause when the drag starts. A label follows the
  • pointer and Esc cancels.

  • The Cards layout is gone. Everyone uses List + details.
  • Success messages are solid green so you notice them.

ERD / Visual Schema

  • Refresh from Database no longer removes tables you've just drawn. A table that was never created is
  • marked as a draft, with a link to create it. Only tables actually dropped from the database are offered
    for removal, unticked by default. The Apply button says what it will do.

  • Tables removed with a card's βœ• no longer come back after reopening the diagram (SOZO-1684).

AI assistants (MCP)

  • AI assistants can create charts on SQL connections, and diagram edits keep primary keys however
  • they're given (SOZO-1683, SOZO-1685). A chart only reads: an assistant can't save one that would change
    data.

  • Settings β€Ί MCP Server hides the token in the client configuration until you reveal it. Copying still
  • copies the real token (SOZO-1686).

Everywhere else

  • Scrolling no longer stutters when the workspace autosaves. It waits until you pause.
  • A tab dragged from a popout back into the main window lands where you drop it, including on a panel
  • edge to split it.

  • Connection Manager: Select All ticks every row and switches to Select None.
  • Licence activation: if a licence can't be installed, the app retries, then keeps running on the
  • Community plan. The banner says why and lets you try again.

Read the full v1.0.2664 release notes

v1.0.2689-mongodb34

October 31, 2026

VisuaLeaf v1.0.2688-mongodb34 Release Notes

Release Date: November 3, 2026

Legacy build for MongoDB 3.4 – 4.0. Brings in everything from the main app 1.0.2351 – 1.0.2687. For MongoDB 4.2 or newer, use the standard VisuaLeaf build.


Headline: Commit Mode, a Rebuilt Data Transfer, BI-Style Dashboards, and Connection Errors That Say What's Wrong

Edits are safer. A new commit mode for MongoDB collections and SQL tables holds grid edits for review
until you commit them, and the SQL Editor gains manual commit with Commit and Rollback. Batch update and
delete now only touch the rows you picked: an empty "Selected Documents" can no longer hit the whole
collection. Undo reverses a grid edit against the database.

Data Transfer has been rebuilt around a task list with details beside it, a live preview and a **Check
job** button, and copies now keep column types and values exact across all 29 SQL engines, MongoDB and
files. Data Compare works on SQL tables and compares structure as well as rows. **Dashboards work like a
BI tool, with dashboard-wide filters, click-to-filter and a new look for every chart. Parquet, XML and
Excel** join CSV as export formats.

Failed connections tell you why, with the server's real reason and a What to check hint. Also
shipping: a rebuilt Query Profiler for MongoDB and SQL, tab groups, multiple windows, a
faster data grid, SSH host key checking and a redesigned Connection Manager.


MongoDB

  • Batch operations only touch the rows you picked. "Selected Documents" is greyed out when nothing is
  • selected, and selecting a field in the tree view targets only its own document, not every document with
    the same value. The Batch Insert, Update and Delete dialogs were redesigned.

  • Insert a document from a template. Batch Insert's Template option builds a document that passes the
  • collection's validator (required fields, allowed values, ranges, lengths), and opens by default on
    collections with a validator. Without one, the template comes from fields in existing documents.

  • Validation rules on every field. The Schema Builder shows each field's rules as chips, and the data
  • viewer shows a Rules chip for fields on collections with a real validator.

  • Correlated $lookup joins run (MongoDB 3.6 or newer). A $lookup with let and pipeline used to be stripped down before
  • it was sent and rejected by the server. The stage is now sent as written, and the aggregation builder can
    add let variables and a sub-pipeline, with autocomplete from the joined collection.

  • Expressions in the visual query builder, such as comparing one field to another in the same document.
  • Hover an ObjectId to see when it was created, in the MongoDB and SQL grids and trees.
  • Add Documents from an empty collection, next to Import Data and Generate Mock Data. Creating a
  • collection opens it.

  • A type-first Create Collection builder. Each collection type shows only its own settings, with a live
  • preview of the db.createCollection command. Combinations MongoDB would reject can't be built.

  • Connection-level shells. Open a shell without picking a database; the tab follows use. Collections
  • created from the shell show up in the sidebar.

  • Replica Set page built around health: a strip with members up, how many it can lose, oplog window and
  • last election, member cards showing lag, and a side panel with plain-word actions once you **Enable
    changes**.

  • MongoDB Clusters is a full-screen workspace with a topology diagram, real node roles instead of
  • "UNKNOWN", a Collections view showing where each sharded collection's data lives, and a guided flow to
    shard a collection.

  • Cluster tools are easier to find. Connections show a sharded or replica set badge, the
  • right-click menu only offers what works on that server, and a new Database menu holds Sharding,
    Replica Set, Clusters and statistics.

  • Expanding a field in aggregation results expands only that row, $elemMatch queries and JSON Schema
  • validation errors were fixed, and very large whole numbers keep their exact value through a shell-format export and import.

SQL

  • SQL Batch Update works on every engine, and "All rows" says plainly that it affects the whole table.
  • Batch update and delete refuse a request that names no rows, no filter and no "all rows" choice.

  • Manual commit in the SQL Editor. Nothing you run is permanent until Commit; Rollback undoes it.
  • An UNCOMMITTED badge counts waiting statements, with warnings when a transaction sits idle, when a MySQL,
    MariaDB or Oracle statement would commit on its own, and when a PostgreSQL transaction has failed.

  • Every column type saves and shows its real value: bigint, decimal, money, UUIDs, intervals, enums,
  • arrays and JSON keep their values, and dates and times are no longer shifted.

  • Long scripts finish. A 40,000-statement dump that froze the editor now runs in about 15 seconds, with
  • long runs of inserts summarised as one line.

  • A two-row SQL Editor toolbar, scripts that remember their schema, Run enabled on a connection with no
  • database picked, and a result view that is remembered between runs.

  • Stop a running query. Stop or Escape cancels a table-grid query on the server.
  • Explain works on every SQL engine, never runs your statement, and has Plan, SQL and JSON tabs.
  • Table Management edits SQLite, DuckDB and ClickHouse tables. ClickHouse type changes are checked
  • before anything runs. What an engine can't do is greyed out with a reason.

  • Edit tables with no key by nominating the columns that identify a row.
  • Drop Schema from the schema's right-click menu, with the statement shown first.
  • Snowflake is a new connection type. ClickHouse moves to a newer driver: connections no longer stall
  • for 30 seconds, primary keys are reported, and tables sorted by an expression are editable.

  • Schemas everywhere: Tasks, Export, Import, Mock Data, Sequences, Procedures, breadcrumbs and the query
  • builder all use the schema you picked instead of the engine's default.

  • Accurate PostgreSQL row counts in Table Stats. Never-analyzed tables are counted exactly, and very
  • large ones show an estimate as ~N.

  • Create Table, indexes and table edits produce valid SQL on every engine, column edits keep their size,
  • and an empty table or result still shows its columns.

Grid Editing

  • Commit mode for MongoDB and SQL. Edits are held, highlighted and counted until you commit them. A
  • Review dialog lists every change with Commit, Discard and per-row Revert. On MongoDB, a document someone
    else changed in the meantime is refused while the rest commit. On by default for SQL, off for MongoDB.

  • Undo (Ctrl/⌘-Z) reverses a cell edit, insert or delete against the database.
  • Up to 6Γ— faster scrolling on wide tables, with no blank grid during fast scrolling. The tree view and
  • row and column resizing are smooth too.

  • Expand a JSON column recursively, and Import and Export buttons now sit in the grid toolbar.

Data Transfer & Data Sync

What you can see

  • Rebuilt around "List + details". Tasks are listed on the left with what's running, failed and up
  • next at the top, with nested folders, task tabs, multi-select, and drag and drop of tasks and jobs.

  • Edit jobs in the page with a live preview of what a job will write, and Check job to test the
  • mapping and every connection without writing anything.

  • Failed tasks say why in plain words, and Monitor and History were redesigned with run history,
  • success rates and duration charts.

What you can do

  • Copies keep every type and value on all 29 SQL engines, MongoDB and files: same-family copies keep
  • the source's own types, numbers stay exact, dates and times aren't shifted, ENUM, SET and JSON columns
    survive, and primary and foreign keys are carried over.

  • Dumps restore exactly, and restored tables keep counting from the right id.
  • MongoDB β†’ SQL samples 1,000 documents to pick column types, so later fields are kept and 12.5 is no
  • longer stored as 12.

  • No silent mistakes. A copy into a missing database no longer writes into the default one, "Replace
  • all" no longer leaves you without a table, rejected rows show as errors, and row counts are real.

  • Tasks start in a fraction of a second, and the page stays responsive with 200+ tasks.
  • Data Sync has three steps, can pick several collections or tables at once, explains why a job can't
  • be created, and SQL β†’ SQL sync works.

Data Compare

  • SQL tables can be compared and synced, alongside MongoDB pairs in one session.
  • Compare structure, not just rows: columns, keys, indexes, checks and triggers for SQL, and indexes,
  • validator and options for MongoDB, with a live script preview of what Sync would run.

  • Sync in either direction. Removals start unticked and the dialog says so; Make identical ticks
  • everything. You're warned about data that would break a change.

  • Match two whole databases in one step, or right-click a database β†’ Compare Structure with….
  • Syncs no longer fail with "Connection is not available", and a 20,000-row sync takes about half a second.

Exports & Imports

  • Export to Parquet, XML and Excel from any table, view, query, database, collection or aggregation.
  • Excel gets real cells and spills onto further sheets; XML keeps nested structure; Parquet keeps types.

  • CSV exports let you choose the separator and header row, and no longer shift SQL dates.
  • MongoDB exports keep fields missing from the first document, and SQL Server imports keep Unicode.

Dashboards & Charts

  • Dashboard-wide filters and click-to-filter. A filter narrows each chart instead of overriding it.
  • Build charts from fields by dropping them into Group by, Measure, Color and Filters, with Show me
  • suggestions. Older charts open in the boxes when they map exactly, with Keep my query instead.

  • A new look for every chart in light and dark mode, a redesigned editor, real pivot grids and tree
  • charts, and cards you can resize from any edge.

  • SQL charts run their query live, so dashboard filters reach them.

Visual Schema (ERD)

  • Refresh from Database no longer removes tables you've just drawn, and removed tables stay removed.
  • Export as SVG, comments on tables and columns, Preview Data from a card, and JSON references.
  • Materialize can add missing columns and foreign keys to existing tables, and lets you pick the schema.
  • A reopened diagram keeps its engine and schema, and required columns are marked with \*.

Connections

  • The real reason a connection failed, with a What to check hint, on every MongoDB, SQL and Redis
  • error. Test Connection makes one login attempt and shows the server's own reason.

  • MySQL, MariaDB and TiDB servers that require encryption are reconnected with TLS automatically, and
  • the app tells you. A TLS mode you picked yourself is never overridden.

  • Pasting a SQL connection URL keeps its TLS settings and names any part of the URL it couldn't use.
  • A redesigned Connection Manager: a folder tree of projects and environments, a details pane,
  • file-browser selection and drag-to-move.

  • Certificate files for PostgreSQL and MySQL, full certificate checking through SSH tunnels, Neon over SSH,
  • and Redis over TLS with your own certificates all work. Connection strings with spaces connect.

Workspace & UI

  • Tab groups by connection, database or by hand, with names, colours, collapse and pop-out, plus
  • preview tabs and new tab menu actions (Clone, Split, Close to the left/right).

  • Multiple windows (⌘/Ctrl+Shift+N), each with its own workspace, reopened at the next launch.
  • Settings rebuilt into clear sections, with adjustable text size.
  • Customizable sidebars: reorder or hide tools, pin up to 10 shortcuts, and create databases, schemas,
  • collections and tables from the sidebar.

  • A SQL Query Profiler on par with the MongoDB one, which was rebuilt to group queries by shape, flag
  • collection scans and suggest indexes. Both have charts, filters and index advice.

  • Guided tours for each feature, a Tutorials list, and a clearer Getting Started window.
  • An Inbox for announcements from VisuaLeaf, and in-app surveys with a reward wheel.
  • Scrolling no longer stutters on autosave, and theme changes reach every window at once.

AI Assistants (MCP)

  • Let an AI assistant drive the app over MCP, from Settings β€Ί MCP Server, with a status-bar switch and
  • separate permission groups. It can open panels, read schemas and rows, build pipelines, manage tasks,
    work with ER diagrams and charts, and set up comparisons.

  • Writing actions need their own switch, such as applying a comparison or creating a connection
  • ("Connections β€” create" in Settings β€Ί MCP). Charts it creates only read.

  • MCP is a licensed feature, and your plan applies to the assistant too. The token stays hidden until
  • you reveal it.

Licensing

  • Free plan limits: up to 5 saved JSON schemas and 3 ER diagrams. The cap applies only when creating
  • new ones; what you already have is kept.

  • Community Edition works again after a plan or trial ends, Settings is always reachable, and the
  • expired dialog appears once instead of hourly. Changing your clock no longer costs you your licence.

  • Trial changes apply without signing out, and Refresh License picks up a plan bought online.
  • Activation works behind corporate proxies. If a licence can't be installed, the app retries, then runs
  • on the Community plan and lets you try again.

Desktop App

  • Quitting waits for a full shutdown, so reopening right away works, and clicking the app while it
  • closes reopens it afterwards.

  • Faster, lighter startup: the window appears about 1.5 seconds sooner, desktop storage is compacted,
  • and broken windows reload themselves. A leftover backend from a crash is stopped automatically.

  • No empty console window flashes on Windows at first launch.
  • Crash reports are opt-in under Settings β€Ί Diagnostics.

Security

  • SSH tunnels check the server's identity. A changed host key is refused before your password or key is
  • sent. The SSH tab shows the trusted key and has Forget trusted key.

  • MongoDB TLS checks the server's host name with "Trust my own CA" or the OS trust store.
  • The desktop backend listens only on your own machine, and markup in search highlights can no longer
  • run as HTML.


Notes

  • MongoDB TLS: if a connection stops working after upgrading, its certificate was issued for a different
  • host name. Connect with the name on the certificate, or switch to "Accept any certificate".

  • SSH: if a server you tunnel through is rebuilt, use Forget trusted key on the SSH tab, then
  • connect again.

  • Commit mode is on by default for SQL tables. Turn it off from the table toolbar to save each edit
  • straight away.

  • Installing: close VisuaLeaf 3.4 before running the installer for this version.
  • ClickHouse connections download a newer driver on first connect. Snowflake connections pin a
  • compatibility option in their connection string; leave it in place.

  • "Task Manager" is now "Data Transfer", and write modes were renamed (Upsert β†’ Add or update, Insert β†’
  • Add new only, Drop & Insert β†’ Replace all). Saved jobs are unchanged.

  • SQL-to-SQL tasks created in earlier versions may have every column set to text. Open the task, click
  • Generate Field Mappings, and save.

  • Treat your MCP token like a password. Anyone holding it can drive that window.
  • Query history keeps the 10,000 most recent unstarred entries. Live Data Sync from MySQL 8.4 or newer is
  • not supported yet.

Read the full v1.0.2689-mongodb34 release notes

v1.0.2639

October 29, 2026

VisuaLeaf v1.0.2630 Release Notes (+ v1.0.2639 Quick Fixes)

Release Date: October 27 2026 Β· Quick fixes October 28 2026

Covers 1.0.2513 – 1.0.2639.


Headline: Compare and Sync Structure, a Rebuilt Data Transfer, Commit Mode for MongoDB and SQL, and Many Windows

Data Compare now compares structure as well as rows. It shows which columns, keys, indexes, checks and
triggers differ between two tables or collections, and can make the target match in either direction.
Data Transfer has been rebuilt around a list of tasks with the selected task's details beside it, and
the job editor now opens in the page with a live preview. A new commit mode for MongoDB and SQL holds
your grid edits for review until you commit them, and the SQL Editor gains manual commit with Commit and
Rollback. **VisuaLeaf
also opens in more than one window**, each with its own workspace.

Also shipping: a SQL Query Profiler on par with the MongoDB one, **a full-screen MongoDB Clusters
workspace with a new sharding view, an Inbox for announcements from us, a type-first Create
Collection builder, SQL schemas in the Schema Manager, a refreshed Redis browser, ENUM and SET
columns that survive every copy, and activation behind corporate proxies**.


Data Compare: Structure, Not Just Rows

What you can see

  • A Structure tab next to Data. For each pair it lists what differs: columns, primary key, indexes,
  • foreign keys, CHECK constraints and triggers for SQL tables, and indexes, validator and options for
    MongoDB collections. Tables or collections that exist on only one side are listed too.

  • One flat list, coloured by what Sync would do. Additions, removals and changes each have their own
  • colour, the changed words are highlighted, and clicking an item explains the difference. Turn on
    Show identical to see everything.

  • A live script preview shows the exact statements Sync would run for the items you've ticked. You
  • can copy it.

  • Find the pairs that differ at a glance. Each pair shows its last result ("12 modified Β· 3 missing",
  • βœ“ identical, Failed). Differing pairs are listed first, and the tabs show counts such as "3 differ"
    or "Identical".

  • Renamed indexes and constraints aren't flagged. Items are matched by what they do, not by name.

What you can do

  • Sync structure in either direction. Pick source β†’ target or target β†’ source and tick what to
  • apply. Additions and changes start ticked; removals and triggers start unticked. Before you confirm,
    the dialog warns about data that would break a change (duplicates, NULLs, rows that fail a new CHECK).

  • Choose Data, Structure or Both next to Run. The choice is saved with the session.
  • Right-click a database or schema β†’ Compare Structure with… to pick a target. Tables and collections
  • are paired by name and the compare runs straight away.

  • Change one side of a pair without rebuilding it, and match two whole databases in one step.
  • Compare tables with no primary key. The target's key is used to match rows.
  • Saved canvases survive a restart, including whole-database cards.

Fixes

  • Comparing or syncing many pairs on one connection no longer fails with "Connection is not available".
  • Pairs wait their turn instead, and one connection always stays free for browsing. If a connection's
    pool is too small for a pair, Run says so before it starts.

  • Row sync writes every column type back on every SQL engine and MongoDB, including PostgreSQL enums. A
  • 20,000-row sync finishes in about half a second instead of four minutes.

  • A stopped or restarted sync no longer re-syncs rows it had already finished.
  • The progress dialog completes instead of freezing at 0%, and the summary says why rows failed, not
  • just how many.


Data Transfer, Rebuilt

What you can see

  • "List + details" is the new default layout. Tasks are listed on the left, one slim row each, with
  • "Running now", "Failed last run" and "Up next" at the top. The selected task fills the right side.

  • Folders form a nested tree. Subfolders sit under their parents, and a folded folder still shows
  • when something inside it failed.

  • Task tabs. Open several tasks in tabs above the details pane. A single click opens a preview tab;
  • double-clicking or editing keeps it.

  • Every task header uses the same layout: status, schedule, parallel runs, last run and its duration,
  • a progress bar for the current run, and run history underneath.

  • Failed tasks say why in plain words, such as a missing table, an unreachable database, a rejected
  • login or invalid filter JSON, instead of "N items failed".

  • Monitor and History, redesigned. Monitor puts every running, paused, queued, finished and upcoming
  • run in one list, ordered by urgency, with details beside it. History shows each task's last 12 runs,
    its success rate and a duration chart, and lists what fails most and which runs are slowest.

What you can do

  • Select several tasks with Ctrl/⌘-click or Shift-click, then run, move, export or delete them
  • together, or drag them all onto a folder.

  • Create tasks and folders from the list, drag tasks between folders, and rename folders in place.
  • Favourites and colours work in both layouts.

  • Move a job to another task from the job's own menu.
  • Edit jobs in the page. The job editor opens over the task list with a breadcrumb back, so your
  • place in the list is kept.

  • Live preview shows what the job will write for a real source row, and updates as you change the
  • mapping.

  • Check job reads sample rows, runs your mapping and tests every connection without writing
  • anything. It lists what's missing or failing, with links to fix each item.

  • Quick actions add a masked field, a value from a lookup or a computed field in one click.
  • Plainer labels: "Add or update", "Add new only" and "Replace all"; "Write null", "Skip field" and
  • "Use default".

  • CSV exports let you choose the separator and whether the first row is column names.
  • Database-to-database SQL copies and SQL dumps keep primary and foreign keys, with a checkbox to
  • turn this off.

  • Pick the schema for a whole-database SQL target.

Faster

  • With 200+ tasks, a click anywhere used to freeze the app for about 2.5 seconds. It now takes about
  • a tenth of a second, and a job's editor opens in about 150 ms.

  • The mapping grid redraws about 7Γ— faster on wide tables, and hovering over it no longer slows the
  • rest of the app.

Copies you can trust, on all 29 SQL engines

  • ENUM and SET columns stay ENUM and SET in copies, native and portable dumps and restores, and arrive
  • as text on engines without enums.

  • Unsigned integers keep their range.
  • No silent writes to the wrong place. A copy into a missing database no longer writes into the
  • connection's default database, and a SQL Server restore no longer ignores the target schema.

  • No silent row loss on CrateDB. Rows it refuses now show up as errors instead of disappearing under
  • "Completed".

  • Fixed: DuckDB restores into a schema could delete the source table; copies of SQLite JSON, H2/HSQLDB
  • UUIDs, ClickHouse fixed-length strings and DuckDB times; QuestDB decimals; SQLite copies within one
    file ("database is locked").


Data Sync

  • Three steps instead of four: Source β†’ What to sync β†’ Review.
  • Pick several collections or tables at once with Select all. Each becomes a rule copying to the
  • same name.

  • Says why a job can't be created, for example a live mode on a source that can't stream changes, or
  • a target sync can't write to. A link takes you to the fix.

  • SQL β†’ SQL sync works, and row estimates work for SQL sources.
  • Fixed jobs from SQL Server, Oracle and CrateDB-family sources, and MongoDB jobs that failed after a
  • collection was dropped and reloaded.


Commit Mode for MongoDB and SQL

Review edits before saving

  • Commit mode for MongoDB collections and SQL tables. Edits you make in the table and tree views are
  • held until you commit them, instead of being written straight away.

  • Pending changes are highlighted (changed, added, removed) and stay through paging, sorting and
  • refreshing. A "N pending" chip in the toolbar shows how many are waiting.

  • A Review dialog lists every change field by field, with Commit, Discard and per-row Revert. Undo
  • drops the latest pending edit without writing anything.

  • On MongoDB, nothing is overwritten by accident. If someone else changed a document after you
  • started editing it, that document is refused, and the rest still commit.

  • A change that fails stays pending with its reason, so you can fix it and try again. Closing a tab with
  • pending changes asks first.

  • Turn it on or off from the toolbar. It's remembered separately for MongoDB and SQL: **on by default for
  • SQL tables, off by default for MongoDB**.

Manual commit in the SQL Editor

  • Switch a tab from Auto-commit to Manual commit. Nothing you run is permanent until you press
  • Commit; Rollback undoes it all.

  • An UNCOMMITTED badge shows how many statements are waiting and how old they are. You're warned
  • after 10 minutes idle, when MySQL, MariaDB or Oracle commits on its own because of a schema change, and
    when an error aborts a PostgreSQL transaction.

  • Closing a tab with an open transaction asks first. The mode is remembered per tab.

Stop a running query

  • Stop (or Escape) cancels a table-grid query on the server. Running again replaces the previous
  • query, and closing the tab cancels it.


Multiple Windows

  • Window β†’ New Window (⌘/Ctrl+Shift+N) opens a full VisuaLeaf window with its own unsaved workspace.
  • Open in New Window in the Workspace Manager opens a saved one.

  • One workspace per window. Two windows never overwrite each other's workspace, and a workspace open
  • elsewhere is marked and brings that window to the front.

  • Closing a window with unsaved tabs asks Save / Discard / Cancel. Quitting gives you one summary for
  • every window.

  • Your windows reopen at the next launch.
  • Popped-out tabs return to the window they came from, and disconnecting warns you if another window is
  • using the connection.


SQL Query Profiler

The SQL profiler now works like the MongoDB one.

What you can see

  • A chart over time, one line per statement, plus a table of statements with each engine's own
  • statistics: rows examined vs. returned, scans without an index, cache hit ratio, p95, CPU and lock time.

  • A detail panel for each statement, with Statement, Statistics, Explain, Index advice and Executions.
  • Index advice suggests a CREATE INDEX in your engine's dialect, and recognises indexes that
  • already cover the statement.

  • ClickHouse profiling, for the first time, including history from before you opened the tab.
  • Honest status. If profiling can't read anything, it says why and names the setting or permission
  • to fix, instead of showing "Profiling on Β· 0 statements".

What you can do

  • Reset the engine's counters where the engine allows it.
  • Resize the chart to make more room for the table.

Fixes

  • SQL Server statements are no longer cut off at 200 characters, and Explain works on SQL Server.
  • First/last-seen times are no longer shifted by hours.

MongoDB Clusters & Sharding

What you can see

  • Clusters is a full-screen workspace, with your clusters on the left and the selected one on the
  • right, organised into Overview, Sharding and Settings.

  • A topology diagram of routers, shards and config servers. Click a node to see its details.
  • Real node roles. Sharded clusters show their primaries, secondaries, arbiters and routers instead
  • of "UNKNOWN".

  • A new Collections view for sharding. Each sharded collection shows which shard holds its data,
  • in plain words ("All on shard0 – the balancer is off"), and unsharded collections are listed largest
    first.

What you can do

  • Shard a collection with a guided flow: pick the collection, pick a key from the Shard Key
  • Advisor's suggestions, then review the index check, what the balancer will do and the exact command.

  • Turn balancing on or off, and change, move or unshard a collection from the side panel.
  • Compound index and shard keys are shown in their real order. They were sometimes shown alphabetised.

Schemas, Tables & Collections

Schema Manager

  • One New Schema menu for MongoDB and SQL: start from scratch, from an existing collection or table,
  • or from a JSON Schema file.

  • SQL Schema Editor. Design a table with the full Create Table form, add personal notes, save it,
  • export it as .sql, or open it in Create Table.

  • Import a live SQL table as a saved schema, with its columns, sizes, defaults, keys, foreign keys
  • and indexes.

  • SQL schemas are grouped as connection β†’ database β†’ schema β†’ table, with engine logos.

Create Collection

  • A type-first builder: Standard, Capped, Time series, View or Clustered. Each type shows only its
  • own settings, and a live preview shows the exact db.createCollection command.

  • Combinations MongoDB would reject can't be built, and the Create button says what's still missing.
  • The collation, storage engine, write concern and JSON Schema dialogs have been restyled and work in
  • light mode.

Create Table & Table Management

  • Create flows confirm and take you to what you made. After creating a table, database, schema or
  • collection you get a short "created" message, and the new item is selected and briefly highlighted in
    the sidebar. A new table opens in the same tab.

  • Types that take a length start with a real, editable value (VARCHAR 255, DECIMAL 10,2).
  • Restyled Create Table, trigger and CHECK forms, and Storage tab.
  • Column edits keep their size on every engine. Making a column NOT NULL no longer shrinks
  • NVARCHAR(255) to NVARCHAR(1) on SQL Server, and column edits on CockroachDB no longer fail.

  • Fixed Drop Trigger on MySQL/MariaDB, ClickHouse tables with JSON or Array columns, Oracle column
  • edits, and comments on SQL Server, Oracle, DuckDB, H2 and HSQLDB.

  • ENUM and SET columns show their real type in the grid and the sidebar.

Visual Schema

  • Export as SVG.
  • Preview Data straight from a table card.
  • Materialize can add missing columns and foreign keys to tables that already exist, instead of
  • skipping them. The preview shows the plan for each table.

  • Dragging a foreign key follows what each engine accepts. INTEGER links to int4, and pairs the engine
  • would refuse are blocked, with an explanation.

  • Table cards show each column type as your engine writes it, with its size.

Sidebar

  • Refreshes after schema changes made in Materialize, Table Management or Events, and Refresh now
  • picks up new columns on tables you've already expanded.

  • Drag connections onto projects and environments to organise them.
  • SQL breadcrumbs are clickable, like the MongoDB ones.

Improvements

Inbox

  • Announcements from VisuaLeaf, newest first, with an unread badge on the right sidebar. Open it from
  • the sidebar, ⌘K, or Help β†’ Inbox. Important announcements pop up once, after any startup prompts.

Query Builders

  • SQL joins: a searchable table picker that lists related tables first.
  • MongoDB aggregation: more window operators in $setWindowFields (moving averages, derivatives,
  • integrals, covariance, fill by interpolation and more), and steadier editing of the pipeline.

Redis

  • Restyled key browser, CLI and server views.
  • Save searches and scripts, and turn auto-refresh on or off. Binary values are marked as not
  • editable as text.

Data Grid

  • Generate Mock Data from an empty collection or table.
  • The table's right edge now runs to the last row and is the right colour in light mode.
  • Dragging the splitters in the SQL table view is smooth. They used to lag or not move.

Export & Import

  • Int64 values above 2^53 survive a MongoDB shell-format export and import exactly.
  • PostgreSQL array columns in SQL Script exports restore correctly.

Connections

  • Clearer message for old MongoDB servers: "This server is MongoDB 4.0. VisuaLeaf supports MongoDB
  • 4.2 to 9.0…", instead of a driver error.

  • Better support for Azure Cosmos DB and Azure DocumentDB.

Startup

  • The window appears about 1.5 seconds sooner after the backend is ready.
  • Smaller, faster desktop storage. Stored data is compacted in the background, so a store that had
  • grown to hundreds of MB shrinks to tens of MB, and the app uses much less memory.

  • No more black or blank windows. If part of the app fails to load at startup, the window reloads
  • itself. If a saved workspace can't be restored, a new one opens and the broken one is left untouched.

  • If a leftover VisuaLeaf backend from an earlier crash is still running, it's stopped automatically.
  • No empty console window flashes on Windows at first launch.

Also in This Release (v1.0.2622 – v1.0.2628)

Data Compare

  • Sync tells you what it leaves out. Sync never removes anything unless you tick it, so it could
  • report one change next to fourteen differences. A note now says so: "13 removals not ticked β€” they
    would drop 11 indexes, 1 column and 1 foreign key from market", with Tick them too.

  • Make identical ticks every difference, removals included, and opens the usual review. Triggers are
  • still left for you to tick.

  • When a sync would remove something, the review says so at the top, and the button reads **Run,
  • including removals**.

Data Transfer

  • "Replace all" into a SQL table no longer leaves you without a table. The new table is created with
  • your database's own column types, and on PostgreSQL, SQL Server and Redshift the old table is kept if
    the new one can't be created.

  • Timestamps with a time zone keep it in the copy.
  • The live preview shows "If missing β†’ use default" exactly as the run writes it. It used to show
  • an empty value.

  • A computed value can use another computed value defined above it.

SQL

  • Review before saving now works on the SQL tab. Edits to a query on one table are held for review,
  • just like the table view. Before, they were saved straight away even with Review ON. When an edit
    can't be held (a join, or rows without their key columns), you're told it was saved.

  • The SQL Editor remembers your result view. Switch a result to Table and the next run opens in
  • Table, not the default.

  • Breadcrumbs show the schema: connection β€Ί database β€Ί schema β€Ί table. Click the schema to open its
  • Tables Info. Clicking the table finds it in its own schema, so two tables with the same name in
    different schemas are no longer mixed up.

MongoDB

  • Add Documents from an empty collection, next to Import Data and Generate Mock Data. Paste one
  • document or an array of them.

  • Creating a collection opens it. The Create Collection tab becomes the new collection's tab, as
  • Create Table does for SQL.

AI Assistants (MCP)

  • An assistant can set up a comparison for you: several tables or collections in one session, opened
  • on the Structure step with the differences already checked.

  • An assistant can add and connect a connection for MongoDB, DocumentDB, Cosmos DB and every SQL
  • engine, with the same checks as the New Connection form. It's **off until you allow "Connections β€”
    create" in Settings β€Ί MCP**. Until then, assistants fill in the form for you to save, as before.


Quick Fixes in v1.0.2639

Finding Cluster and Sharding Tools

  • Connections say what they are. A connected MongoDB connection shows a sharded or replica set
  • badge. Click it to open Sharding or Replica Set.

  • The right-click menu only offers what works. A standalone server no longer offers Sharding or
  • Replica Set, which only led to a "not available" message, and a replica set no longer offers Sharding.

  • A new Database menu in the menu bar: Sharding, Replica Set, Manage MongoDB Clusters, Database
  • Statistics and Collections Statistics.

  • Clusters VisuaLeaf runs are marked "managed". If VisuaLeaf deployed a cluster, or you adopted it
  • with Find untracked (Docker or SSH hosts), its connection gets Manage Cluster…, which opens MongoDB
    Clusters on that cluster. Its Sharding panel links there too, for starting new shards, resizing or
    upgrading.

Explain

  • Explain works on every SQL engine, from the Query Profiler and the table view. It used to fail on
  • profiled statements with placeholder values on PostgreSQL-family engines and SQL Server, and on
    CockroachDB, CrateDB, QuestDB, RisingWave, Materialize, H2 and HSQLDB.

  • Explain never runs your statement. On DuckDB it used to: explaining a CREATE TABLE created the
  • table.

  • Plans are easier to read. Oracle plans keep every step, so joins show as a tree again. Engines that
  • return their plan as text show it as an aligned tree instead of a block of JSON.

Visual Schema

  • JSON references. Draw a relation to a field inside a JSON column. It's shown dashed and saved with
  • the diagram, and never becomes a real foreign key when you materialize or export. (SOZO-1666)

  • Pick the schema when materializing a blank diagram on engines that have schemas. It always went to
  • public before. (SOZO-1676)

  • A reopened diagram keeps its engine and schema. Diagrams drawn from a non-default schema used to
  • come back pointing at public, and Refresh offered to delete every card.

  • After editing DBML, Refresh from Database no longer reports "Foreign keys differ" on every table.

Data Transfer

  • JSON columns stay JSON in database-to-database copies: JSONB on PostgreSQL-family engines, JSON on
  • MySQL, MariaDB, TiDB and DuckDB. They used to arrive as plain text.

  • Row counts are real. A job that copied nothing reported "100 rows". It now reports 0.

Connections

  • Redis over TLS with your own certificates. Choose the system's certificate authorities or paste your
  • own, and add a client certificate and key if the server asks for one. Servers with a private CA can now
    connect.

  • MongoDB TLS checks the server's host name with "Trust my own CA" or the OS trust store, so a
  • certificate issued to a different host is refused. See Notes.

Also

  • Create Collection offers the "constraint" validation level (MongoDB 9.0+) and the "errorAndLog"
  • action (MongoDB 8.1+). Options your server is too old for are greyed out with the version they need.
    (SOZO-1668)

  • DuckDB JSON values in the SQL grid show as values (true, 42, "text") instead of "{ 6 fields }".
  • SQL table breadcrumbs stay on one line in narrow tabs.
  • The aggregation builder's menus were restyled.

Notes

  • MongoDB TLS connections now check the server's host name when set to "Trust my own CA" or the OS
  • trust store. If a connection stops working after upgrading, its certificate was issued for a different
    host name: connect with the name on the certificate, or switch the connection to "Accept any
    certificate". SSH-tunnelled connections are unchanged.

  • Commit mode is on by default for SQL tables. Edits in the SQL grid are held until you press
  • Commit. Turn it off from the table toolbar if you'd rather save each edit straight away. On MongoDB it
    starts off; turn it on from the collection toolbar.

  • Data Transfer opens in "List + details" unless you had already chosen Cards.
  • Write-mode labels were renamed (Upsert β†’ Add or update, Insert β†’ Add new only, Drop & Insert β†’ Replace
  • all). Saved jobs are unchanged.

  • Query history on desktop now keeps the 10,000 most recent unstarred entries. Starred entries are
  • always kept.

  • Live Data Sync from MySQL 8.4 or newer is reported as not ready, because the bundled change
  • connector doesn't support it yet.

Read the full v1.0.2639 release notes

v1.0.2630

October 28, 2026

VisuaLeaf v1.0.2630 Release Notes - Quick Addition to Version 2621

Release Date: October 27 2026

Covers 1.0.2513 – 1.0.2630.


Headline: Compare and Sync Structure, a Rebuilt Data Transfer, Commit Mode for MongoDB and SQL, and Many Windows

Data Compare now compares structure as well as rows. It shows which columns, keys, indexes, checks and
triggers differ between two tables or collections, and can make the target match in either direction.
Data Transfer has been rebuilt around a list of tasks with the selected task's details beside it, and
the job editor now opens in the page with a live preview. A new commit mode for MongoDB and SQL holds
your grid edits for review until you commit them, and the SQL Editor gains manual commit with Commit and
Rollback. **VisuaLeaf
also opens in more than one window**, each with its own workspace.

Also shipping: a SQL Query Profiler on par with the MongoDB one, **a full-screen MongoDB Clusters
workspace with a new sharding view, an Inbox for announcements from us, a type-first Create
Collection builder, SQL schemas in the Schema Manager, a refreshed Redis browser, ENUM and SET
columns that survive every copy, and activation behind corporate proxies**.


Data Compare: Structure, Not Just Rows

What you can see

  • A Structure tab next to Data. For each pair it lists what differs: columns, primary key, indexes,
  • foreign keys, CHECK constraints and triggers for SQL tables, and indexes, validator and options for
    MongoDB collections. Tables or collections that exist on only one side are listed too.

  • One flat list, coloured by what Sync would do. Additions, removals and changes each have their own
  • colour, the changed words are highlighted, and clicking an item explains the difference. Turn on
    Show identical to see everything.

  • A live script preview shows the exact statements Sync would run for the items you've ticked. You
  • can copy it.

  • Find the pairs that differ at a glance. Each pair shows its last result ("12 modified Β· 3 missing",
  • βœ“ identical, Failed). Differing pairs are listed first, and the tabs show counts such as "3 differ"
    or "Identical".

  • Renamed indexes and constraints aren't flagged. Items are matched by what they do, not by name.

What you can do

  • Sync structure in either direction. Pick source β†’ target or target β†’ source and tick what to
  • apply. Additions and changes start ticked; removals and triggers start unticked. Before you confirm,
    the dialog warns about data that would break a change (duplicates, NULLs, rows that fail a new CHECK).

  • Choose Data, Structure or Both next to Run. The choice is saved with the session.
  • Right-click a database or schema β†’ Compare Structure with… to pick a target. Tables and collections
  • are paired by name and the compare runs straight away.

  • Change one side of a pair without rebuilding it, and match two whole databases in one step.
  • Compare tables with no primary key. The target's key is used to match rows.
  • Saved canvases survive a restart, including whole-database cards.

Fixes

  • Comparing or syncing many pairs on one connection no longer fails with "Connection is not available".
  • Pairs wait their turn instead, and one connection always stays free for browsing. If a connection's
    pool is too small for a pair, Run says so before it starts.

  • Row sync writes every column type back on every SQL engine and MongoDB, including PostgreSQL enums. A
  • 20,000-row sync finishes in about half a second instead of four minutes.

  • A stopped or restarted sync no longer re-syncs rows it had already finished.
  • The progress dialog completes instead of freezing at 0%, and the summary says why rows failed, not
  • just how many.


Data Transfer, Rebuilt

What you can see

  • "List + details" is the new default layout. Tasks are listed on the left, one slim row each, with
  • "Running now", "Failed last run" and "Up next" at the top. The selected task fills the right side.

  • Folders form a nested tree. Subfolders sit under their parents, and a folded folder still shows
  • when something inside it failed.

  • Task tabs. Open several tasks in tabs above the details pane. A single click opens a preview tab;
  • double-clicking or editing keeps it.

  • Every task header uses the same layout: status, schedule, parallel runs, last run and its duration,
  • a progress bar for the current run, and run history underneath.

  • Failed tasks say why in plain words, such as a missing table, an unreachable database, a rejected
  • login or invalid filter JSON, instead of "N items failed".

  • Monitor and History, redesigned. Monitor puts every running, paused, queued, finished and upcoming
  • run in one list, ordered by urgency, with details beside it. History shows each task's last 12 runs,
    its success rate and a duration chart, and lists what fails most and which runs are slowest.

What you can do

  • Select several tasks with Ctrl/⌘-click or Shift-click, then run, move, export or delete them
  • together, or drag them all onto a folder.

  • Create tasks and folders from the list, drag tasks between folders, and rename folders in place.
  • Favourites and colours work in both layouts.

  • Move a job to another task from the job's own menu.
  • Edit jobs in the page. The job editor opens over the task list with a breadcrumb back, so your
  • place in the list is kept.

  • Live preview shows what the job will write for a real source row, and updates as you change the
  • mapping.

  • Check job reads sample rows, runs your mapping and tests every connection without writing
  • anything. It lists what's missing or failing, with links to fix each item.

  • Quick actions add a masked field, a value from a lookup or a computed field in one click.
  • Plainer labels: "Add or update", "Add new only" and "Replace all"; "Write null", "Skip field" and
  • "Use default".

  • CSV exports let you choose the separator and whether the first row is column names.
  • Database-to-database SQL copies and SQL dumps keep primary and foreign keys, with a checkbox to
  • turn this off.

  • Pick the schema for a whole-database SQL target.

Faster

  • With 200+ tasks, a click anywhere used to freeze the app for about 2.5 seconds. It now takes about
  • a tenth of a second, and a job's editor opens in about 150 ms.

  • The mapping grid redraws about 7Γ— faster on wide tables, and hovering over it no longer slows the
  • rest of the app.

Copies you can trust, on all 29 SQL engines

  • ENUM and SET columns stay ENUM and SET in copies, native and portable dumps and restores, and arrive
  • as text on engines without enums.

  • Unsigned integers keep their range.
  • No silent writes to the wrong place. A copy into a missing database no longer writes into the
  • connection's default database, and a SQL Server restore no longer ignores the target schema.

  • No silent row loss on CrateDB. Rows it refuses now show up as errors instead of disappearing under
  • "Completed".

  • Fixed: DuckDB restores into a schema could delete the source table; copies of SQLite JSON, H2/HSQLDB
  • UUIDs, ClickHouse fixed-length strings and DuckDB times; QuestDB decimals; SQLite copies within one
    file ("database is locked").


Data Sync

  • Three steps instead of four: Source β†’ What to sync β†’ Review.
  • Pick several collections or tables at once with Select all. Each becomes a rule copying to the
  • same name.

  • Says why a job can't be created, for example a live mode on a source that can't stream changes, or
  • a target sync can't write to. A link takes you to the fix.

  • SQL β†’ SQL sync works, and row estimates work for SQL sources.
  • Fixed jobs from SQL Server, Oracle and CrateDB-family sources, and MongoDB jobs that failed after a
  • collection was dropped and reloaded.


Commit Mode for MongoDB and SQL

Review edits before saving

  • Commit mode for MongoDB collections and SQL tables. Edits you make in the table and tree views are
  • held until you commit them, instead of being written straight away.

  • Pending changes are highlighted (changed, added, removed) and stay through paging, sorting and
  • refreshing. A "N pending" chip in the toolbar shows how many are waiting.

  • A Review dialog lists every change field by field, with Commit, Discard and per-row Revert. Undo
  • drops the latest pending edit without writing anything.

  • On MongoDB, nothing is overwritten by accident. If someone else changed a document after you
  • started editing it, that document is refused, and the rest still commit.

  • A change that fails stays pending with its reason, so you can fix it and try again. Closing a tab with
  • pending changes asks first.

  • Turn it on or off from the toolbar. It's remembered separately for MongoDB and SQL: **on by default for
  • SQL tables, off by default for MongoDB**.

Manual commit in the SQL Editor

  • Switch a tab from Auto-commit to Manual commit. Nothing you run is permanent until you press
  • Commit; Rollback undoes it all.

  • An UNCOMMITTED badge shows how many statements are waiting and how old they are. You're warned
  • after 10 minutes idle, when MySQL, MariaDB or Oracle commits on its own because of a schema change, and
    when an error aborts a PostgreSQL transaction.

  • Closing a tab with an open transaction asks first. The mode is remembered per tab.

Stop a running query

  • Stop (or Escape) cancels a table-grid query on the server. Running again replaces the previous
  • query, and closing the tab cancels it.


Multiple Windows

  • Window β†’ New Window (⌘/Ctrl+Shift+N) opens a full VisuaLeaf window with its own unsaved workspace.
  • Open in New Window in the Workspace Manager opens a saved one.

  • One workspace per window. Two windows never overwrite each other's workspace, and a workspace open
  • elsewhere is marked and brings that window to the front.

  • Closing a window with unsaved tabs asks Save / Discard / Cancel. Quitting gives you one summary for
  • every window.

  • Your windows reopen at the next launch.
  • Popped-out tabs return to the window they came from, and disconnecting warns you if another window is
  • using the connection.


SQL Query Profiler

The SQL profiler now works like the MongoDB one.

What you can see

  • A chart over time, one line per statement, plus a table of statements with each engine's own
  • statistics: rows examined vs. returned, scans without an index, cache hit ratio, p95, CPU and lock time.

  • A detail panel for each statement, with Statement, Statistics, Explain, Index advice and Executions.
  • Index advice suggests a CREATE INDEX in your engine's dialect, and recognises indexes that
  • already cover the statement.

  • ClickHouse profiling, for the first time, including history from before you opened the tab.
  • Honest status. If profiling can't read anything, it says why and names the setting or permission
  • to fix, instead of showing "Profiling on Β· 0 statements".

What you can do

  • Reset the engine's counters where the engine allows it.
  • Resize the chart to make more room for the table.

Fixes

  • SQL Server statements are no longer cut off at 200 characters, and Explain works on SQL Server.
  • First/last-seen times are no longer shifted by hours.

MongoDB Clusters & Sharding

What you can see

  • Clusters is a full-screen workspace, with your clusters on the left and the selected one on the
  • right, organised into Overview, Sharding and Settings.

  • A topology diagram of routers, shards and config servers. Click a node to see its details.
  • Real node roles. Sharded clusters show their primaries, secondaries, arbiters and routers instead
  • of "UNKNOWN".

  • A new Collections view for sharding. Each sharded collection shows which shard holds its data,
  • in plain words ("All on shard0 – the balancer is off"), and unsharded collections are listed largest
    first.

What you can do

  • Shard a collection with a guided flow: pick the collection, pick a key from the Shard Key
  • Advisor's suggestions, then review the index check, what the balancer will do and the exact command.

  • Turn balancing on or off, and change, move or unshard a collection from the side panel.
  • Compound index and shard keys are shown in their real order. They were sometimes shown alphabetised.

Schemas, Tables & Collections

Schema Manager

  • One New Schema menu for MongoDB and SQL: start from scratch, from an existing collection or table,
  • or from a JSON Schema file.

  • SQL Schema Editor. Design a table with the full Create Table form, add personal notes, save it,
  • export it as .sql, or open it in Create Table.

  • Import a live SQL table as a saved schema, with its columns, sizes, defaults, keys, foreign keys
  • and indexes.

  • SQL schemas are grouped as connection β†’ database β†’ schema β†’ table, with engine logos.

Create Collection

  • A type-first builder: Standard, Capped, Time series, View or Clustered. Each type shows only its
  • own settings, and a live preview shows the exact db.createCollection command.

  • Combinations MongoDB would reject can't be built, and the Create button says what's still missing.
  • The collation, storage engine, write concern and JSON Schema dialogs have been restyled and work in
  • light mode.

Create Table & Table Management

  • Create flows confirm and take you to what you made. After creating a table, database, schema or
  • collection you get a short "created" message, and the new item is selected and briefly highlighted in
    the sidebar. A new table opens in the same tab.

  • Types that take a length start with a real, editable value (VARCHAR 255, DECIMAL 10,2).
  • Restyled Create Table, trigger and CHECK forms, and Storage tab.
  • Column edits keep their size on every engine. Making a column NOT NULL no longer shrinks
  • NVARCHAR(255) to NVARCHAR(1) on SQL Server, and column edits on CockroachDB no longer fail.

  • Fixed Drop Trigger on MySQL/MariaDB, ClickHouse tables with JSON or Array columns, Oracle column
  • edits, and comments on SQL Server, Oracle, DuckDB, H2 and HSQLDB.

  • ENUM and SET columns show their real type in the grid and the sidebar.

Visual Schema

  • Export as SVG.
  • Preview Data straight from a table card.
  • Materialize can add missing columns and foreign keys to tables that already exist, instead of
  • skipping them. The preview shows the plan for each table.

  • Dragging a foreign key follows what each engine accepts. INTEGER links to int4, and pairs the engine
  • would refuse are blocked, with an explanation.

  • Table cards show each column type as your engine writes it, with its size.

Sidebar

  • Refreshes after schema changes made in Materialize, Table Management or Events, and Refresh now
  • picks up new columns on tables you've already expanded.

  • Drag connections onto projects and environments to organise them.
  • SQL breadcrumbs are clickable, like the MongoDB ones.

Improvements

Inbox

  • Announcements from VisuaLeaf, newest first, with an unread badge on the right sidebar. Open it from
  • the sidebar, ⌘K, or Help β†’ Inbox. Important announcements pop up once, after any startup prompts.

Query Builders

  • SQL joins: a searchable table picker that lists related tables first.
  • MongoDB aggregation: more window operators in $setWindowFields (moving averages, derivatives,
  • integrals, covariance, fill by interpolation and more), and steadier editing of the pipeline.

Redis

  • Restyled key browser, CLI and server views.
  • Save searches and scripts, and turn auto-refresh on or off. Binary values are marked as not
  • editable as text.

Data Grid

  • Generate Mock Data from an empty collection or table.
  • The table's right edge now runs to the last row and is the right colour in light mode.
  • Dragging the splitters in the SQL table view is smooth. They used to lag or not move.

Export & Import

  • Int64 values above 2^53 survive a MongoDB shell-format export and import exactly.
  • PostgreSQL array columns in SQL Script exports restore correctly.

Connections

  • Clearer message for old MongoDB servers: "This server is MongoDB 4.0. VisuaLeaf supports MongoDB
  • 4.2 to 9.0…", instead of a driver error.

  • Better support for Azure Cosmos DB and Azure DocumentDB.

Startup

  • The window appears about 1.5 seconds sooner after the backend is ready.
  • Smaller, faster desktop storage. Stored data is compacted in the background, so a store that had
  • grown to hundreds of MB shrinks to tens of MB, and the app uses much less memory.

  • No more black or blank windows. If part of the app fails to load at startup, the window reloads
  • itself. If a saved workspace can't be restored, a new one opens and the broken one is left untouched.

  • If a leftover VisuaLeaf backend from an earlier crash is still running, it's stopped automatically.
  • No empty console window flashes on Windows at first launch.

Also in This Release (v1.0.2622 – v1.0.2628)

Data Compare

  • Sync tells you what it leaves out. Sync never removes anything unless you tick it, so it could
  • report one change next to fourteen differences. A note now says so: "13 removals not ticked β€” they
    would drop 11 indexes, 1 column and 1 foreign key from market", with Tick them too.

  • Make identical ticks every difference, removals included, and opens the usual review. Triggers are
  • still left for you to tick.

  • When a sync would remove something, the review says so at the top, and the button reads **Run,
  • including removals**.

Data Transfer

  • "Replace all" into a SQL table no longer leaves you without a table. The new table is created with
  • your database's own column types, and on PostgreSQL, SQL Server and Redshift the old table is kept if
    the new one can't be created.

  • Timestamps with a time zone keep it in the copy.
  • The live preview shows "If missing β†’ use default" exactly as the run writes it. It used to show
  • an empty value.

  • A computed value can use another computed value defined above it.

SQL

  • Review before saving now works on the SQL tab. Edits to a query on one table are held for review,
  • just like the table view. Before, they were saved straight away even with Review ON. When an edit
    can't be held (a join, or rows without their key columns), you're told it was saved.

  • The SQL Editor remembers your result view. Switch a result to Table and the next run opens in
  • Table, not the default.

  • Breadcrumbs show the schema: connection β€Ί database β€Ί schema β€Ί table. Click the schema to open its
  • Tables Info. Clicking the table finds it in its own schema, so two tables with the same name in
    different schemas are no longer mixed up.

MongoDB

  • Add Documents from an empty collection, next to Import Data and Generate Mock Data. Paste one
  • document or an array of them.

  • Creating a collection opens it. The Create Collection tab becomes the new collection's tab, as
  • Create Table does for SQL.

AI Assistants (MCP)

  • An assistant can set up a comparison for you: several tables or collections in one session, opened
  • on the Structure step with the differences already checked.

  • An assistant can add and connect a connection for MongoDB, DocumentDB, Cosmos DB and every SQL
  • engine, with the same checks as the New Connection form. It's **off until you allow "Connections β€”
    create" in Settings β€Ί MCP**. Until then, assistants fill in the form for you to save, as before.


Notes

  • Commit mode is on by default for SQL tables. Edits in the SQL grid are held until you press
  • Commit. Turn it off from the table toolbar if you'd rather save each edit straight away. On MongoDB it
    starts off; turn it on from the collection toolbar.

  • Data Transfer opens in "List + details" unless you had already chosen Cards.
  • Write-mode labels were renamed (Upsert β†’ Add or update, Insert β†’ Add new only, Drop & Insert β†’ Replace
  • all). Saved jobs are unchanged.

  • Query history on desktop now keeps the 10,000 most recent unstarred entries. Starred entries are
  • always kept.

  • Live Data Sync from MySQL 8.4 or newer is reported as not ready, because the bundled change
  • connector doesn't support it yet.

Read the full v1.0.2630 release notes

v1.0.2621

October 27, 2026

VisuaLeaf v1.0.2621 Release Notes

Release Date: October 27 2026

Covers 1.0.2513 – 1.0.2621.


Headline: Compare and Sync Structure, a Rebuilt Data Transfer, Commit Mode for MongoDB and SQL, and Many Windows

Data Compare now compares structure as well as rows. It shows which columns, keys, indexes, checks and
triggers differ between two tables or collections, and can make the target match in either direction.
Data Transfer has been rebuilt around a list of tasks with the selected task's details beside it, and
the job editor now opens in the page with a live preview. A new commit mode for MongoDB and SQL holds
your grid edits for review until you commit them, and the SQL Editor gains manual commit with Commit and
Rollback. **VisuaLeaf
also opens in more than one window**, each with its own workspace.

Also shipping: a SQL Query Profiler on par with the MongoDB one, **a full-screen MongoDB Clusters
workspace with a new sharding view, an Inbox for announcements from us, a type-first Create
Collection builder, SQL schemas in the Schema Manager, a refreshed Redis browser, ENUM and SET
columns that survive every copy, and activation behind corporate proxies**.


Data Compare: Structure, Not Just Rows

What you can see

  • A Structure tab next to Data. For each pair it lists what differs: columns, primary key, indexes,
  • foreign keys, CHECK constraints and triggers for SQL tables, and indexes, validator and options for
    MongoDB collections. Tables or collections that exist on only one side are listed too.

  • One flat list, coloured by what Sync would do. Additions, removals and changes each have their own
  • colour, the changed words are highlighted, and clicking an item explains the difference. Turn on
    Show identical to see everything.

  • A live script preview shows the exact statements Sync would run for the items you've ticked. You
  • can copy it.

  • Find the pairs that differ at a glance. Each pair shows its last result ("12 modified Β· 3 missing",
  • βœ“ identical, Failed). Differing pairs are listed first, and the tabs show counts such as "3 differ"
    or "Identical".

  • Renamed indexes and constraints aren't flagged. Items are matched by what they do, not by name.

What you can do

  • Sync structure in either direction. Pick source β†’ target or target β†’ source and tick what to
  • apply. Additions and changes start ticked; removals and triggers start unticked. Before you confirm,
    the dialog warns about data that would break a change (duplicates, NULLs, rows that fail a new CHECK).

  • Choose Data, Structure or Both next to Run. The choice is saved with the session.
  • Right-click a database or schema β†’ Compare Structure with… to pick a target. Tables and collections
  • are paired by name and the compare runs straight away.

  • Change one side of a pair without rebuilding it, and match two whole databases in one step.
  • Compare tables with no primary key. The target's key is used to match rows.
  • Saved canvases survive a restart, including whole-database cards.

Fixes

  • Comparing or syncing many pairs on one connection no longer fails with "Connection is not available".
  • Pairs wait their turn instead, and one connection always stays free for browsing. If a connection's
    pool is too small for a pair, Run says so before it starts.

  • Row sync writes every column type back on every SQL engine and MongoDB, including PostgreSQL enums. A
  • 20,000-row sync finishes in about half a second instead of four minutes.

  • A stopped or restarted sync no longer re-syncs rows it had already finished.
  • The progress dialog completes instead of freezing at 0%, and the summary says why rows failed, not
  • just how many.


Data Transfer, Rebuilt

What you can see

  • "List + details" is the new default layout. Tasks are listed on the left, one slim row each, with
  • "Running now", "Failed last run" and "Up next" at the top. The selected task fills the right side.

  • Folders form a nested tree. Subfolders sit under their parents, and a folded folder still shows
  • when something inside it failed.

  • Task tabs. Open several tasks in tabs above the details pane. A single click opens a preview tab;
  • double-clicking or editing keeps it.

  • Every task header uses the same layout: status, schedule, parallel runs, last run and its duration,
  • a progress bar for the current run, and run history underneath.

  • Failed tasks say why in plain words, such as a missing table, an unreachable database, a rejected
  • login or invalid filter JSON, instead of "N items failed".

  • Monitor and History, redesigned. Monitor puts every running, paused, queued, finished and upcoming
  • run in one list, ordered by urgency, with details beside it. History shows each task's last 12 runs,
    its success rate and a duration chart, and lists what fails most and which runs are slowest.

What you can do

  • Select several tasks with Ctrl/⌘-click or Shift-click, then run, move, export or delete them
  • together, or drag them all onto a folder.

  • Create tasks and folders from the list, drag tasks between folders, and rename folders in place.
  • Favourites and colours work in both layouts.

  • Move a job to another task from the job's own menu.
  • Edit jobs in the page. The job editor opens over the task list with a breadcrumb back, so your
  • place in the list is kept.

  • Live preview shows what the job will write for a real source row, and updates as you change the
  • mapping.

  • Check job reads sample rows, runs your mapping and tests every connection without writing
  • anything. It lists what's missing or failing, with links to fix each item.

  • Quick actions add a masked field, a value from a lookup or a computed field in one click.
  • Plainer labels: "Add or update", "Add new only" and "Replace all"; "Write null", "Skip field" and
  • "Use default".

  • CSV exports let you choose the separator and whether the first row is column names.
  • Database-to-database SQL copies and SQL dumps keep primary and foreign keys, with a checkbox to
  • turn this off.

  • Pick the schema for a whole-database SQL target.

Faster

  • With 200+ tasks, a click anywhere used to freeze the app for about 2.5 seconds. It now takes about
  • a tenth of a second, and a job's editor opens in about 150 ms.

  • The mapping grid redraws about 7Γ— faster on wide tables, and hovering over it no longer slows the
  • rest of the app.

Copies you can trust, on all 29 SQL engines

  • ENUM and SET columns stay ENUM and SET in copies, native and portable dumps and restores, and arrive
  • as text on engines without enums.

  • Unsigned integers keep their range.
  • No silent writes to the wrong place. A copy into a missing database no longer writes into the
  • connection's default database, and a SQL Server restore no longer ignores the target schema.

  • No silent row loss on CrateDB. Rows it refuses now show up as errors instead of disappearing under
  • "Completed".

  • Fixed: DuckDB restores into a schema could delete the source table; copies of SQLite JSON, H2/HSQLDB
  • UUIDs, ClickHouse fixed-length strings and DuckDB times; QuestDB decimals; SQLite copies within one
    file ("database is locked").


Data Sync

  • Three steps instead of four: Source β†’ What to sync β†’ Review.
  • Pick several collections or tables at once with Select all. Each becomes a rule copying to the
  • same name.

  • Says why a job can't be created, for example a live mode on a source that can't stream changes, or
  • a target sync can't write to. A link takes you to the fix.

  • SQL β†’ SQL sync works, and row estimates work for SQL sources.
  • Fixed jobs from SQL Server, Oracle and CrateDB-family sources, and MongoDB jobs that failed after a
  • collection was dropped and reloaded.


Commit Mode for MongoDB and SQL

Review edits before saving

  • Commit mode for MongoDB collections and SQL tables. Edits you make in the table and tree views are
  • held until you commit them, instead of being written straight away.

  • Pending changes are highlighted (changed, added, removed) and stay through paging, sorting and
  • refreshing. A "N pending" chip in the toolbar shows how many are waiting.

  • A Review dialog lists every change field by field, with Commit, Discard and per-row Revert. Undo
  • drops the latest pending edit without writing anything.

  • On MongoDB, nothing is overwritten by accident. If someone else changed a document after you
  • started editing it, that document is refused, and the rest still commit.

  • A change that fails stays pending with its reason, so you can fix it and try again. Closing a tab with
  • pending changes asks first.

  • Turn it on or off from the toolbar. It's remembered separately for MongoDB and SQL: **on by default for
  • SQL tables, off by default for MongoDB**.

Manual commit in the SQL Editor

  • Switch a tab from Auto-commit to Manual commit. Nothing you run is permanent until you press
  • Commit; Rollback undoes it all.

  • An UNCOMMITTED badge shows how many statements are waiting and how old they are. You're warned
  • after 10 minutes idle, when MySQL, MariaDB or Oracle commits on its own because of a schema change, and
    when an error aborts a PostgreSQL transaction.

  • Closing a tab with an open transaction asks first. The mode is remembered per tab.

Stop a running query

  • Stop (or Escape) cancels a table-grid query on the server. Running again replaces the previous
  • query, and closing the tab cancels it.


Multiple Windows

  • Window β†’ New Window (⌘/Ctrl+Shift+N) opens a full VisuaLeaf window with its own unsaved workspace.
  • Open in New Window in the Workspace Manager opens a saved one.

  • One workspace per window. Two windows never overwrite each other's workspace, and a workspace open
  • elsewhere is marked and brings that window to the front.

  • Closing a window with unsaved tabs asks Save / Discard / Cancel. Quitting gives you one summary for
  • every window.

  • Your windows reopen at the next launch.
  • Popped-out tabs return to the window they came from, and disconnecting warns you if another window is
  • using the connection.


SQL Query Profiler

The SQL profiler now works like the MongoDB one.

What you can see

  • A chart over time, one line per statement, plus a table of statements with each engine's own
  • statistics: rows examined vs. returned, scans without an index, cache hit ratio, p95, CPU and lock time.

  • A detail panel for each statement, with Statement, Statistics, Explain, Index advice and Executions.
  • Index advice suggests a CREATE INDEX in your engine's dialect, and recognises indexes that
  • already cover the statement.

  • ClickHouse profiling, for the first time, including history from before you opened the tab.
  • Honest status. If profiling can't read anything, it says why and names the setting or permission
  • to fix, instead of showing "Profiling on Β· 0 statements".

What you can do

  • Reset the engine's counters where the engine allows it.
  • Resize the chart to make more room for the table.

Fixes

  • SQL Server statements are no longer cut off at 200 characters, and Explain works on SQL Server.
  • First/last-seen times are no longer shifted by hours.

MongoDB Clusters & Sharding

What you can see

  • Clusters is a full-screen workspace, with your clusters on the left and the selected one on the
  • right, organised into Overview, Sharding and Settings.

  • A topology diagram of routers, shards and config servers. Click a node to see its details.
  • Real node roles. Sharded clusters show their primaries, secondaries, arbiters and routers instead
  • of "UNKNOWN".

  • A new Collections view for sharding. Each sharded collection shows which shard holds its data,
  • in plain words ("All on shard0 – the balancer is off"), and unsharded collections are listed largest
    first.

What you can do

  • Shard a collection with a guided flow: pick the collection, pick a key from the Shard Key
  • Advisor's suggestions, then review the index check, what the balancer will do and the exact command.

  • Turn balancing on or off, and change, move or unshard a collection from the side panel.
  • Compound index and shard keys are shown in their real order. They were sometimes shown alphabetised.

Schemas, Tables & Collections

Schema Manager

  • One New Schema menu for MongoDB and SQL: start from scratch, from an existing collection or table,
  • or from a JSON Schema file.

  • SQL Schema Editor. Design a table with the full Create Table form, add personal notes, save it,
  • export it as .sql, or open it in Create Table.

  • Import a live SQL table as a saved schema, with its columns, sizes, defaults, keys, foreign keys
  • and indexes.

  • SQL schemas are grouped as connection β†’ database β†’ schema β†’ table, with engine logos.

Create Collection

  • A type-first builder: Standard, Capped, Time series, View or Clustered. Each type shows only its
  • own settings, and a live preview shows the exact db.createCollection command.

  • Combinations MongoDB would reject can't be built, and the Create button says what's still missing.
  • The collation, storage engine, write concern and JSON Schema dialogs have been restyled and work in
  • light mode.

Create Table & Table Management

  • Create flows confirm and take you to what you made. After creating a table, database, schema or
  • collection you get a short "created" message, and the new item is selected and briefly highlighted in
    the sidebar. A new table opens in the same tab.

  • Types that take a length start with a real, editable value (VARCHAR 255, DECIMAL 10,2).
  • Restyled Create Table, trigger and CHECK forms, and Storage tab.
  • Column edits keep their size on every engine. Making a column NOT NULL no longer shrinks
  • NVARCHAR(255) to NVARCHAR(1) on SQL Server, and column edits on CockroachDB no longer fail.

  • Fixed Drop Trigger on MySQL/MariaDB, ClickHouse tables with JSON or Array columns, Oracle column
  • edits, and comments on SQL Server, Oracle, DuckDB, H2 and HSQLDB.

  • ENUM and SET columns show their real type in the grid and the sidebar.

Visual Schema

  • Export as SVG.
  • Preview Data straight from a table card.
  • Materialize can add missing columns and foreign keys to tables that already exist, instead of
  • skipping them. The preview shows the plan for each table.

  • Dragging a foreign key follows what each engine accepts. INTEGER links to int4, and pairs the engine
  • would refuse are blocked, with an explanation.

  • Table cards show each column type as your engine writes it, with its size.

Sidebar

  • Refreshes after schema changes made in Materialize, Table Management or Events, and Refresh now
  • picks up new columns on tables you've already expanded.

  • Drag connections onto projects and environments to organise them.
  • SQL breadcrumbs are clickable, like the MongoDB ones.

Improvements

Inbox

  • Announcements from VisuaLeaf, newest first, with an unread badge on the right sidebar. Open it from
  • the sidebar, ⌘K, or Help β†’ Inbox. Important announcements pop up once, after any startup prompts.

Query Builders

  • SQL joins: a searchable table picker that lists related tables first.
  • MongoDB aggregation: more window operators in $setWindowFields (moving averages, derivatives,
  • integrals, covariance, fill by interpolation and more), and steadier editing of the pipeline.

Redis

  • Restyled key browser, CLI and server views.
  • Save searches and scripts, and turn auto-refresh on or off. Binary values are marked as not
  • editable as text.

Data Grid

  • Generate Mock Data from an empty collection or table.
  • The table's right edge now runs to the last row and is the right colour in light mode.
  • Dragging the splitters in the SQL table view is smooth. They used to lag or not move.

Export & Import

  • Int64 values above 2^53 survive a MongoDB shell-format export and import exactly.
  • PostgreSQL array columns in SQL Script exports restore correctly.

Connections

  • Clearer message for old MongoDB servers: "This server is MongoDB 4.0. VisuaLeaf supports MongoDB
  • 4.2 to 9.0…", instead of a driver error.

  • Better support for Azure Cosmos DB and Azure DocumentDB.

Startup

  • The window appears about 1.5 seconds sooner after the backend is ready.
  • Smaller, faster desktop storage. Stored data is compacted in the background, so a store that had
  • grown to hundreds of MB shrinks to tens of MB, and the app uses much less memory.

  • No more black or blank windows. If part of the app fails to load at startup, the window reloads
  • itself. If a saved workspace can't be restored, a new one opens and the broken one is left untouched.

  • If a leftover VisuaLeaf backend from an earlier crash is still running, it's stopped automatically.
  • No empty console window flashes on Windows at first launch.

Notes

  • Commit mode is on by default for SQL tables. Edits in the SQL grid are held until you press
  • Commit. Turn it off from the table toolbar if you'd rather save each edit straight away. On MongoDB it
    starts off; turn it on from the collection toolbar.

  • Data Transfer opens in "List + details" unless you had already chosen Cards.
  • Write-mode labels were renamed (Upsert β†’ Add or update, Insert β†’ Add new only, Drop & Insert β†’ Replace
  • all). Saved jobs are unchanged.

  • Query history on desktop now keeps the 10,000 most recent unstarred entries. Starred entries are
  • always kept.

  • Live Data Sync from MySQL 8.4 or newer is reported as not ready, because the bundled change
  • connector doesn't support it yet.

Read the full v1.0.2621 release notes

Looking for the latest version?

Go to Downloads