Free to use · No sign-up

Local SQL workbench

Query CSV and Parquet files, join local tables and export results with DuckDB in your browser.

CSV / Parquet tables · up to 5 files, 20 MB combinedSQL query

DuckDB runs locally. Use SELECT / WITH, joins and aggregates; no semicolons. Example: SELECT a.* FROM table1 a JOIN table2 b ON a.id = b.id LIMIT 100. CSV headers and types are inferred. Up to 5,000 result rows / 2 million characters. Engine loading counts toward the 30-second deadline.

What it does

Query CSV and Parquet files, join local tables and export results with DuckDB in your browser.

How to use it

Import up to five CSV/Parquet files, 20 MB combined. Files become table1, table2 and so on.
Write one SELECT or WITH query, without semicolons. Run SQL and export JSON or CSV.
Processing stays in this browser. No sign-up or upload.

Example

SELECT COUNT(*) AS total FROM table1 returns the imported row count.

Methodology

DuckDB WebAssembly imports files into in-memory tables in a worker. External access is disabled before user queries.

Limitations

30-second deadline including engine loading, 128 MB engine memory setting, 5,000 output rows and 2 million output characters. CSV types are inferred; CSV formula cells are escaped. No uploads or persistent database.

Automation

This tool runs in your browser. No upload API or MCP endpoint is available.

View capabilities and documentation

Reference updated 2026-09-17

Join the conversation

Helpful tips, questions, and ideas for making this tool better.

Loading discussion…