Because it does exist, but it's not free. Jetbrains Datagrip has been around for a decade and has connectors for most database archetypes (mongo, sql, redis, duckdb, etc).
Biggest limitation right now is its lack of support for vector style databases like Lance, qdrant, etc.
Datagrip is fantastic. You can also get the same functionality if you buy the IntelliJ IDEA Ultimate Edition, which includes many of the languages and tools that are also separately productized. I recommend this because the database tools integrate so well with the editor. For example, you can have a scratch file open that contains multiple SQL snippets. You can hit cmd-Enter when inside one snippet, and it will execute it and render the results either as a panel or inline inside the editor, notebook-style. Plus, full schema validation and autocompletion in the editor, and support for many, many databases including Postgres and ClickHouse.
Not only that, but the SQL support works for embedded strings in programs written in other languages such as Go. So it knows that some statement conn.Exec("SELECT ...") is SQL and syntax-highlights it, performs schema validation and autocompletion inside the string literal. Not only that, but you can open the string literal as a separate editor and edit it, including doing things like "reformat", which was an unexpected delight when I discovered it.
It's this kind of "feature stacking", which features working organically with each other, that makes Jetbrains IDEs so damn good.
But the basic database tools are also superb. Its table view is really fast. It has syntax highlighting (e.g. if a column value is JSON), live editing (including the ability to open a column value as an editor, in which case you get all the usual syntax tools), and even graph rendering with support for multiple data series and grouping in a single graph.
There is also excellent support for exporting data. You can mark a bunch of result rows and copy them as CSV or as SQL INSERT statements, or you can save the entire result to a file. This is how I often export data from BigQuery, as it's much more convenient than Google's own tooling (the web UI is particularly bad, requiring that you export the query result to a GCS bucket first).
These database UIs aren't technically difficult to do. But somehow nobody else seems capable. The closest I can think of is Microsoft's tooling around SQL Server, which is pretty slick, albeit MSSQL-specific. I often wonder how Jetbrains, which is a pretty small company, can be so effective and produce such an incredibly feature-rich product portfolio.
Missed that - I am a little unclear how being wrapped in electron as an app is necessarily an advantage - I guess it could be valuable if it were pulled out electron, and you could host it as a service.
Yeah you can definitely do that, I referenced how we do it in another comment. The electron app and running it locally allows you to use TCP protocol which isn't available directly in the browser.
It's mostly just a nicely bundled way to run it if you aren't very technical but still want a easy to use database client.
You can also run commands like this to connect to your database if you want:
Biggest limitation right now is its lack of support for vector style databases like Lance, qdrant, etc.