SQL Workbench
Query local Parquet, CSV and JSON files with real SQL, powered by DuckDB compiled to WebAssembly.
Join files, aggregate columns and export results — without your data ever leaving the browser.
Loading tool…
How it works
The workbench runs DuckDB-WASM inside a Web Worker in your browser. Dropped files are registered with DuckDB by reference — they are read on demand, not copied — and you can query them by filename: SELECT * FROM 'sales.parquet' JOIN 'users.csv' USING (id). The full DuckDB SQL dialect is available, including window functions, aggregates and JSON functions.
Tips & shortcuts
- Press ⌘ Enter (or Ctrl+Enter) to run the query.
- Click a registered file's name to insert it at the cursor, or use its Preview / Schema / Stats buttons for instant one-click queries.
- Join across formats freely — e.g.
SELECT * FROM 'a.parquet' JOIN 'b.csv' USING (id) - Your last query is restored when you come back.