Cleaning messy CSVs and Excel files usually means bouncing between Excel, Jupyter, and df.head() prints just to catch typos like "New Yrok", mixed dates (2025-01-05 vs 12-Jan-2025), $1,250.50 strings, or "true" / "yes" / "false" booleans.
I built PandaScrub (~440 KB VS Code extension + web studio) so you can inspect and clean datasets visually inside VS Code while generating real, editable Pandas / Polars Python code side-by-side.
What it solves in ~30 seconds
Instant Profiling & Anomaly Detection
Open any.csv,.xlsx,.parquet,.json,.sqlite,.duckdb, or PythonDataFramescript right inside VS Code and immediately see null %, unique counts, data types, and zero-variance / mixed-format alerts.-
One-Click Visual Cleaning (No Boilerplate)
-
Fuzzy Deduplicate & Typo Merge: Clusters variants like
"New Yrok","New Yorkk","New Yrk"->"New York". -
Date & Currency Parsing: Turns
$1,250.50into clean floats and standardizes mixed date formats (YYYY-MM-DD,DD/MM/YYYY, etc.). -
Formula & Window Builder: Create calculated columns (
lifetime_spend * 1.20), conditionalIF/THENrules, orGroupBy/ running totals visually. -
Boolean Cleanup: Normalizes mixed
"true","True","yes","false","no","1","0"cleanly.
-
Fuzzy Deduplicate & Typo Merge: Clusters variants like
Diff Changes + 2-Way Python Sync
Toggle Diff Changes to highlight every modified cell against the original dataset, then click Insert Pipeline to inject clean Pandas or Polars code straight into your.pyfile. Edit Python and pressCtrl+S-> the table hot-reloads live.
Try It
-
VS Code Marketplace: Search
PandaScrubin VS Code Extensions - Web Studio: pandascrub.dev
-
Community & Issues:
What's the most annoying data-cleaning step you keep repeating in Pandas? Happy to hear feedback!

Top comments (1)
need you support to make this product better and useful for devs