What it is
When a font lacks a character, the game shows an empty box or a fallback glyph that looks wrong. This tool reads the character map of a TTF, OTF, WOFF or TTC font in your browser and compares it with the text you paste or import: it lists every missing character with its code point, script and a snippet of where it is used, and reports coverage per text and per script. It then builds the minimal character set the text needs, estimates how large the subset font would be and exports it as a BMFont configuration, a pyftsubset command, a CSS unicode-range rule or a JSON report.
When to use it
Run it before shipping a translation to Chinese, Japanese, Korean, Cyrillic or any language with accents, and again whenever strings change. Use it to decide whether one font is enough or a fallback is needed, to cut a 10 MB CJK font down to the few thousand characters you really use, and to feed the exact character list to a bitmap font tool. Text can come from pasted strings or files such as JSON, CSV, PO and properties.
Gotchas & limits
Coverage here means the font maps the character to a glyph; it cannot tell whether the glyph looks right, and complex scripts such as Arabic or Devanagari also need shaping tables that this check does not test. WOFF2 fonts are not read, so convert them to TTF, OTF or WOFF first. The subset size is an estimate before WOFF2 compression, and kerning or ligature tables may shrink by a different amount than assumed. Line breaks, zero-width marks and variation selectors are ignored.