What I work on
Two things: the code that generates the character data, and the writing that explains it. Both sides of this site come from the same place — a preference for checking a claim over repeating one.
The tools are plain browser JavaScript with no backend. Nothing you paste into a generator, detector or remover on this site leaves your machine, because there is no server to send it to. That is an architectural decision, not a policy promise.
How the data gets verified
Every Unicode property published here is generated from the Unicode Character Database, never typed by hand. General Category, White_Space, the NFKC result, UTF-8 byte length and HTML entities are read straight out of UCD 16.0.0 and written into the site's data files by script.
The reason is that this subject is full of confident, repeated mistakes. Two that surfaced while building this site: NFKC folds 16 of the 17 Space Separator characters to an ordinary space, and U+3164 Hangul Filler — the character most guides recommend for blank usernames — folds to U+1160 under NFKC. Neither appears on the competing references I surveyed.
Any property claim on this site can be reproduced in three lines of Python with the standard library's unicodedata module. Checking beats trusting.
What I do not claim to know
Unicode properties are fixed by the standard. Platform behaviour is not. Whether a particular app accepts a particular character this month depends on validation code I cannot see and that changes without announcement.
Where a page describes how a platform behaves, it is compiled from current reporting and should be treated as a starting point — test it in the field before spending a rename token on it. Where a page states a Unicode property, it is generated and exact. Those two kinds of claim are not the same, and I try not to blur them.
Corrections
Corrections are genuinely welcome, especially on platform behaviour. A field report from someone who just tried it beats my best estimate. If something here is wrong or has changed, tell me through the contact page and I will fix it and note the change.
Where to start reading
- The complete invisible character table — all 29 codepoints with the four properties that decide whether each one works
- U+3164 Hangul Filler — the character most often recommended, and the NFKC caveat most guides miss
- The detector — paste any string and see every hidden character named and counted