Query Editor

One Editor, Five Dialects

IntelliShell is a Monaco-based editor with per-engine language support. What you write depends on the connection the tab points at:

• MongoDB: mongo-shell style JavaScript — find, aggregate, CRUD, runCommand, cursors, ObjectId/ISODate literals
• PostgreSQL: multi-statement SQL scripts on a single pinned session — SET and temp tables survive across statements in a run
• SQL Server: T-SQL with proper GO batch handling, the way SSMS does it
• Aerospike: AQL — SELECT, INSERT, UPDATE, DELETE, index DDL, SHOW and STAT
• ClickHouse: native ClickHouse SQL over HTTP — DESCRIBE, PREWHERE and friends

Autocomplete That Knows Your Schema

Suggestions come from your live connection, not a static keyword list: collection and table names, field and column names, MongoDB operators with per-position awareness, and engine-specific SQL keywords. Field suggestions are cached briefly (about five minutes) so the sidebar stays fast. Press Ctrl+Space to trigger suggestions manually.

The list narrows to what belongs where the cursor is. After FROM or JOIN you get table names and nothing else; in a SELECT list or a WHERE clause, the columns of the table that statement is about — including when it is written as public.users or behind an alias. Inside CAST(x AS …) or a CREATE TABLE column you get that dialect's data types, and wherever a value belongs you get its functions, each one inserted with the cursor already between the parentheses.

Each dialect gets its own vocabulary rather than a shared one: 93 keywords and 35 functions for PostgreSQL, 92 and 35 for SQL Server, 116 and 41 for ClickHouse. Every one of those functions is executed against a real server in our test suite, because suggesting a function an engine does not have is worse than suggesting nothing — SQL Server gets CEILING, the others get CEIL. Per-table templates round it off: pick orders: GROUP BY and the whole statement lands, with TOP on SQL Server and LIMIT elsewhere.

On Aerospike the bins of a set are offered with the indexed ones marked, which is not decoration — Aerospike can only filter on an indexed bin, so it decides whether a WHERE runs at all.

Example: MongoDB

db.getCollection("customers").find({
  lifetimeValue: { $gte: 1000 },
  city: "Berlin"
}).sort({ createdAt: -1 }).limit(20)

Example: SQL

SELECT name, email, city
FROM customers
WHERE lifetime_value > 1000
ORDER BY lifetime_value DESC

Three Ways to Run

• Ctrl+Enter — run the selected text
• F5 — run the whole script; each statement gets its own result tab
• F6 — run just the statement under the cursor

How to: iterate on one statement inside a longer script

1. Keep your whole script in the editor — setup statements, the query you're tuning, cleanup
2. Place the cursor inside the statement you're working on and press F6 — only that statement runs
3. Alternatively select any fragment and press Ctrl+Enter to run exactly the selection
4. When everything works, F5 runs the whole script — one result tab per statement

Abort That Actually Aborts

The abort button cancels server-side, not just the connection: pg_cancel_backend on PostgreSQL, request cancellation on SQL Server, and KILL QUERY on ClickHouse — a runaway SELECT * over millions of rows actually stops.

Make It Yours

Font size, tab size, wrapping, line numbers, the minimap and how suggestions behave are all under Options → Editing → Editor, and change every open editor at once. See Settings.

Sensible Limits

Three numbers decide what a statement may cost, and all three are yours to change under Options → Editing → Query:

• Queries default to 50 rows unless you specify a limit yourself (1–10 000) — paging in the results panel fetches more
• Responses are capped at 2 MB per request (1–50 MB); oversized results are truncated with a note that names the limit and where to raise it
• Statements run until they finish unless you set a timeout (up to an hour) — it reaches every engine, SQL Server and Aerospike included

Changes apply to the next run, no restart.

AI Query Generation

Describe what you need in plain language and Sutido generates the query for the current engine, aware of your selected database and schema. Bring your own API key — OpenAI, Google Gemini, or Anthropic Claude — configured in Settings. Requests go directly from your machine to the provider you chose.

Next Steps

Prefer clicking over typing for MongoDB? Try the Query Builder. Then learn what the results panel can do.