Instant profiling for CSV, Parquet and Excel: types, keys, null maps, quality
findings, a SQL workbench and ready-to-run CREATE TABLE DDL.
Everything runs client-side — the file never leaves this tab.
Drop a data file — or browse
.csv.tsv.txt.parquet.xlsx
Orders: 1,340 deliberately messy rows — every quality finding fires. Enrollments: 980 tidy rows with
a unique ID column and a student + course + semester combination to verify in the composite key tester.
▸ Manual import options (auto-detection handles most files)
Private by construction
Profiling runs on DuckDB compiled to WebAssembly, inside this tab. There is no backend and no upload endpoint — pull the network cable after the page loads and everything still works.
Forgiving ingestion
Exotic delimiters, Windows-1252 mojibake, banner lines above the header, malformed rows — auto-handled, and every decision is listed in “How this file was read”.
Real SQL, real DDL
A full DuckDB SQL workbench against your file, plus generated CREATE TABLE statements for SQL Server and Postgres — with verified primary keys.
Profiling…
Reading file
Detecting format & encoding
Starting DuckDB engine
Loading into table
Profiling columns
01
Overview
How this file was read
— every automatic decision, nothing silent
02
Columns
click a row for top values & detail stats
Column
Type
Nulls
Null %
Distinct
Uniqueness
Min → Max
Badges
Composite key tester
Pick two or more columns and verify whether the combination uniquely identifies every row. Verified combinations become primary-key options in the DDL generator.
03
SQL workbench
full DuckDB SQL against table data · Ctrl/⌘+Enter to run
04
Data quality
05
DDL generator
CREATE TABLE from the inferred schema
06
Export
Markdown data dictionary — overview, columns, findings, DDL