Workspace continuity.
Save portable table definitions and SQL tabs together. Import the workspace later with the same query names and order.
VS Code extension · v1.2
SQL for your files.
Inside VS Code.
Right-click a CSV, JSON, Parquet, or S3 file — get a queryable SQL table. Powered by DuckDB. 100% local.
See it in 60 seconds
Query any file
One extension, every format your team keeps sending you.
-
CSV
Auto-detected delimiters, headers, and types.
-
JSON
Nested objects, arrays, and mixed schemas.
-
Parquet
Columnar reads — millions of rows, milliseconds.
-
TSV
Tab-separated files, no config required.
-
JSONL / NDJSON
Streamed logs and events, one row per line.
-
S3
Point at a bucket or prefix — auto-detected region.
Features you'll love
Smart loading.
Local configured tables auto-load as lightweight DuckDB views. S3 stays lazy until you choose Load, with an opt-out for local auto-load.
Complete exports.
Export the full active query to CSV or Parquet from the toolbar,
beyond the visible row limit. Run custom DuckDB
COPY statements too.
Hive partitions, one table.
Query multi-level key=value folders as one table.
Partition values become columns instead of separate tables.
How it works
-
Install
One command in your terminal, or one click in the Marketplace.
$ code --install-extension arunkumar1997.file-sql -
Load a file
Right-click any file in Explorer → Open with File SQL.
Explorer → right-click 'sales.parquet' → Open with File SQL -
Query
Write SQL, press Ctrl+Enter. Results in the panel below.
SELECT * FROM sales LIMIT 10;
See it in your editor
Two ways in, one workflow — no dialogs, no plugins to configure.
s3:// URI or point at a folder outside your
workspace.