IDE Tool Overview
The IDE is a database workbench that runs in the browser while the server does the work: the server holds the connection pools, reads each engine's catalog and runs your queries, and the browser reaches it through the Next.js → Express proxy. Nothing installs on your machine, and no database driver, credential or network path to a database reaches the browser. One editor, navigator and results pane serve relational, document, key-value, wide-column, graph, search and time-series engines, and the navigator shows each engine's own objects rather than a common subset.
Workbench Surfaces
Connection Manager
Your saved connections, held on the server under your account, so they survive sign-out and clearing the browser. Group them with New Group in the + menu. Connections can be imported from the DB DevOps Manager as independent copies (see Connecting to Databases).
Object Explorer / Navigator
A tree of each connected database's objects that loads as you expand it: namespaces, object folders, columns with type, nullability, default and key role, and, on the engines that have them, server-level nodes such as server info, sessions, users, roles and tablespaces. The panel is docked on the right, collapsed to a slim rail until your first connect opens it. The tree sits above an inspector, with a draggable divider between them. The Search objects box at its top finds objects by name once you type two characters. Covered in depth under Schema Explorer.
Query Editor
A Monaco editor with SQL highlighting, dialect-aware formatting and multi-statement execution. Tabs hold separate query texts, and every tab runs against the connection selected in the header. Open tabs and their text are kept in this browser for your account until you log out, so they survive a reload and switching to DB DevOps and back.
Results
A grid for statements that return rows, the affected-row count for INSERT, UPDATE and DELETE, success or the engine's error for DDL, and export to CSV or JSON. Preview says how much a change would affect without keeping it: rows, documents, keys or graph changes depending on the engine.
AI Assistant
A conversational assistant (requires a configured AI provider) grounded in the selected connection's schema. It keeps the conversation, so a follow-up question can build on earlier answers; it sees the chat itself, not the rows a query returned in the grid. Generated queries arrive in code blocks you can insert into the editor or run directly, and a destructive statement is refused either way. Each user's conversation is kept in this browser and restored after a reload, a sign-out and sign-in, and switching between the IDE and DB DevOps; clear it from the panel header. It follows the AI settings: tables an administrator excluded from AI are not part of its schema, and it is refused while AI agents are switched off.
Switching between the IDE and DB DevOps preserves your working state on both sides: the selected connection, open results and in-progress forms are where you left them when you switch back.
Supported Engines
The connection form's Database Type list offers PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, CockroachDB, TimescaleDB, MongoDB, Redis, Apache Cassandra, Neo4j, Elasticsearch, Supabase, PlanetScale, ClickHouse, Amazon DynamoDB, InfluxDB, Google BigQuery and Snowflake. Supabase is PostgreSQL and PlanetScale is MySQL on the wire, so each connects, browses and runs like its engine, including script runs, lint rules and pre-flight checks in DB DevOps; Supabase starts with SSL Mode Require and PlanetScale with Verify Full, since both hosts require TLS. What the navigator shows follows each engine's own catalog:
| Family | Engines | Navigator |
|---|---|---|
| Relational | PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, CockroachDB, Supabase, PlanetScale | Databases or schemas as the engine names them; tables, views, routines, sequences and the engine's other object types; columns, indexes, foreign keys and triggers; users, roles and tablespaces where the engine has them. |
| Time-series | TimescaleDB | The same tree as PostgreSQL. |
| Time-series | InfluxDB | Buckets → measurements; a measurement's columns are _time, its tags and its fields. |
| Document | MongoDB | Databases → collections and views; fields sampled from the first 100 documents, since a collection declares no schema. |
| Key-value | Redis | Key namespaces (the part of each key before its first colon) with how many keys each holds. |
| Wide-column | Apache Cassandra | Keyspaces → tables, materialized views and functions; columns. |
| Graph | Neo4j | Node labels and relationship types, in folders of their own. |
| Search | Elasticsearch | Indices, with their document count, size and health. |
| Key-value | Amazon DynamoDB | Tables, with their approximate item count, size and keys; a table's columns are its partition and sort keys, its index keys and the attributes seen in its first 50 items. |
| Analytical | Google BigQuery | Datasets → tables, views and materialized views, with row counts and sizes; columns with their types and modes. |
| Analytical | Snowflake | Databases → schemas → tables, with row counts and sizes, views and materialized views; columns with their types, nullability and defaults. |
| Analytical | ClickHouse | Databases → tables, views and materialized views; columns. Connects over the HTTP interface (port 8123). |
The per-engine differences are listed under Keyboard Shortcuts & Multi-Database Support.
Layout
- Left, your saved connections and groups. The chevron collapses the rail; drag its edge to resize it, and double-click the edge to reset it.
- Center, the toolbar, the query tabs and the editor. The connection the editor runs against is chosen in the header.
- Below the editor, the results pane; drag its top edge to resize it.
- Right, the navigator: the object tree above the inspector, with a draggable divider (double-click to reset) whose position is kept in this browser.
On a phone-sized screen the registered connections are tucked behind the menu button at the left of the top bar and open as a drawer that closes when you pick a connection, press Esc, or tap outside it, the Object Explorer opens the same way from the button at the right of the top bar, and the editor and results use the full screen width. Tablets and screens narrower than 1280 pixels use the same drawers so the editor keeps its full width, and below 1700 pixels the AI assistant opens over the Object Explorer instead of squeezing the editor.