Free · No signup · Runs in your browser

Profile your CSV. Prove its quality.

Inspect every column, rank deterministic quality findings, test required, type, unique, range, pattern and allowed-value rules, then export a reproducible audit — all in your browser.

01 · How it works

Three steps, then done.

A profile describes what is in a file; a contract proves whether it meets your expectations. This workspace keeps those jobs separate, shows every denominator, and preserves the row-level evidence behind each failure.

i. profile

Load CSV or Excel

Inspect inferred types, completeness, distributions, exact duplicates, mixed values and IQR outliers locally.

ii. contract

Define what must pass

Choose a business key and edit required, type, unique, range, pattern and allowed-value rules.

iii. prove

Export the evidence

Download aggregate HTML or Markdown, full profile statistics, or a disclosed row-level issue ledger.

02 · Why ours

A quality audit that shows its work.

Observed signals stay separate from explicit failures, so a surprising value is not silently called wrong.

  • 01

    Deterministic findings

    Exact duplicates, missingness, mixed value shapes, case variants, concentration and IQR outliers with visible calculations.

  • 02

    Reusable contracts

    Save data-free column rules, arm one before a file loads, and rerun the same checks on the next export.

  • 03

    Business-key evidence

    Test a single or composite key, then see every affected row when keys are blank or repeated.

  • 04

    Hardened exports

    Aggregate reports exclude raw values; the complete issue CSV discloses exactly when failing values are included.

"A profile says what changed. A contract says whether the file is allowed to proceed."
— the audit contract
03 · FAQ

profiler questions.

What is the difference between a profile finding and a contract failure?
A finding describes an observed signal such as missingness, a dominant value, or an IQR outlier. A contract failure means a row broke a rule you explicitly applied, such as required, unique, type, range, pattern, allowed values, or business-key uniqueness.
The profiler uses the IQR rule: values below Q1 minus 1.5 times the interquartile range or above Q3 plus 1.5 times the range are observations for review. They do not fail a contract unless you set an explicit minimum or maximum.
Only its version, name, selected business-key column names, and rule configuration. It never contains filenames, source rows, observed values, profile counts, findings, failures, or audit results.
Standalone HTML and Markdown contain aggregate profile and rule evidence only. Profile CSV contains statistics. Issue CSV is the one raw-value export; it warns before download that row numbers and failing values are included.
Yes. CSV, TXT, XLSX and XLS are read locally in the browser. Multi-sheet workbooks ask you which sheet to audit.
No. Normal profiling, contract validation and report export happen entirely in your browser, and audit results are not stored.