How the gate works

How the CSV Schema & Quality gate works

This tool grades a CSV / data-export file before you trust it. Each signal is scored 0–2; the verdict is the weakest one — then a gate that can only make it worse.

The 7 signals

01Header present

02Header columns unique

03No blank header cells

04Every row same widthtop trigger

05Clean UTF-8 encoding

06All required columns present

07No entirely-empty column

READY

all structural checks pass, columns clean

FIX

a column is dirty or missing

BLOCK

any structural check fails

Structural · worsen-only

Structure is categorical: one failed structural check — a ragged row, a blank header, bad encoding — BLOCKs the whole file no matter how clean the individual columns are. A file that can't line up can't be trusted, regardless of the score.

BLOCK

Why: Every one of six columns reads CLEAN and file quality is a perfect 100 — but one structural check fails: the rows aren't all the same width. A ragged file can't be trusted to line up, so it's BLOCK. Structure is categorical; quality is graduated. weakest signal 0/2 · all six columns CLEAN, file quality 100.

Fix firstEvery row same width — fix the ragged rows so the file parses into a consistent grid before trusting any cell.

A clean sheet didn’t save it. The verdict is the weakest signal — and a top trigger is dispositive on its own — so it names the one thing to fix first. Grades the file structure and column quality, not people. Runs offline in the workbook and engine.

RedHub AIthe AI that tells you when to doubt it redhub.ai/systems/csv-schema-quality-gate