Grid and Review

LocHub's web UI has six tabs: Grid, Review, Glossary, Jobs, Coverage and Inbox. This page covers the first two — the Grid, where you browse and filter every string, and the Review queue, where you work through translations one at a time.

The Grid

The Grid is a virtualized spreadsheet of every localizable string in the project. Each row is one string (a "unit"); columns are Key (sticky), Source (the native-culture text, sticky), and one column per culture you have chosen to show.

Columns

Key and Source are always shown and stay pinned while you scroll sideways.
Every other project culture is available through the Columns button in the toolbar. The culture you are currently working in (the active culture, chosen in the culture picker) is always shown and cannot be hidden; every other culture is a checkbox you can turn on or off. Your choice of extra columns is remembered between sessions.
Clicking a cell for a given culture opens that string in the Review queue's cell panel (see below), scrolled to that specific string.

Grid overview

Cell statuses

Each cell shows the translated text plus a small status badge:

Badge	Meaning
(none)	Empty — nothing translated yet
Draft	An AI draft awaiting review (badge color follows the triage band, below)
Needs fix	The AI draft failed an automatic format check (see Format checks below) or Pull refused it; it does not ship until a human fixes it or approves it anyway
Approved	A human approved the AI draft as-is
Edited	A human edited the AI draft before approving it
Human	The translation was written by a human from scratch
Rejected	A human rejected the draft; it goes back to translation
Outdated	The cell has text, but the source (English) string has changed since it was translated

A cell still loading shows "Loading…"; a cell whose fetch failed renders blank instead of loading forever.

Triage bands (R / Y / G)

Every AI draft is triaged into a band that only affects review priority — it never decides what can ship (see the Settings Reference page for the release policy that controls that):

Red (R) — the draft failed a format check (either kind), was refused, has a judge issue rated major or critical, or (for strings used in UI widgets) an ambiguity the model had to guess at. Review this first.
Yellow (Y) — the model was unsure about context or had to guess (unless the string is UI widget text, which goes to Red instead), or the judge or the deterministic checks raised something minor.
Green (G) — clean: no ambiguity, no judge issues, no precheck issues.

A small, stable share of Green cells (about 4% by default) is still sampled into the Review queue as a blind audit — a spot check on what triage might be missing. An audited cell keeps its Green look in the Grid; only the Review queue treats it specially (see below).

Filters and search

The toolbar above the Grid offers: