Getting started

DataPrep Desk is a browser-based utility for preparing small datasets without installing software or creating an account.

Import a dataset

Open the Data Workbench and choose a .csv, .tsv or .json file, or paste dataset text. A sample dataset is available if you want to try the workflow first. File size is limited to 3 MB, with up to 10,000 data rows and 100 columns.

Supported data formats

CSV and TSV are treated as header-first tables. Quoted values, escaped quotation marks, newlines within quoted CSV values, and CRLF line endings are supported. JSON input must be an array of objects; nested JSON values are represented as JSON text in table cells. Column values remain strings so IDs such as 000125 are not silently converted to numbers.

Inspect quality

The overview reports row and column counts, missing cell counts, exact duplicate rows, inferred column types, and unique non-empty values per column. Type inference is advisory, not a validation guarantee. The workbench previews the first twelve records; downloads contain all rows.

Clean your data

Choose whether to trim whitespace, remove wholly empty rows, deduplicate exact repeated rows, and normalise headers to lowercase snake_case. Click Apply cleaning to preview the changes. You can restore the imported original before exporting. Cleanup does not edit your source file.

Export

Download the current table as CSV, TSV or JSON, or a Markdown quality report containing counts and column names but no raw rows. When exporting text for spreadsheets, cells resembling formulas are prefixed to reduce spreadsheet formula injection risks. JSON exports contain string cell values; dates and numbers are not automatically recast.

JSON utilities

The JSON Tools page can validate, pretty-print and minify a JSON document and download the result. JavaScript JSON parsing can lose precision for very large numerical literals, so use caution with high-precision data.

Privacy & limitations

Parsing, analysis and output generation are performed locally in your browser. The current tools do not upload your dataset to an application server. Hosting providers may still process request metadata when delivering the website. This product is aimed at modest files rather than gigabyte-scale workloads.

Source and feedback

DataPrep Desk is open source. Read the repository or report an issue.