Tidy a description before you paste it into YouTube. It normalises line endings, trims trailing spaces, collapses accidental runs of blank lines while keeping the single blank line that is a real paragraph break, straightens the curly quotes and em dashes a word processor inserts, and removes the invisible characters that survive a trim and then break a search inside your own text. It also strips tracking parameters from your links — including YouTube's own si= parameter, which the share button appends and which identifies the account that did the sharing. Links are rebuilt with the browser's URL parser, so everything except the removed parameters survives byte for byte. Each option can be switched off, and the tool reports exactly what it changed.
| Step | Effect |
|---|---|
| Line endings | Carriage returns removed |
| Invisible characters | Zero-width, soft hyphen, BOM removed |
| Typography | Curly quotes, dashes, ellipsis straightened |
| Link tracking | si, utm_*, fbclid and friends removed |
| Trailing spaces | Trimmed per line |
| Blank lines | Three or more collapse to one |
No. URLs are parsed with the browser's own parser and rebuilt from the parameters that remain, rather than re-encoded — so a link with nothing to strip comes out identical. Trailing punctuation is detached first, so the full stop ending your sentence is not swallowed into the URL.
It is the tracking parameter YouTube's share button appends to every youtu.be link, and it identifies the account that shared it. Leave it in a public description and you hand it to every viewer. The link works exactly the same without it.
No. A single blank line is a deliberate paragraph break and is kept. Only runs of three or more newlines collapse, down to one blank line.
Zero-width spaces, soft hyphens, byte-order marks and directional override characters. They are not whitespace, so a trim leaves them behind, and they are why a search for a phrase in your own description sometimes finds nothing.
5,000 characters, counted here by code point. The counter shows where you stand and flags going over.
Yes, all of them. Space collapsing is off by default, since some descriptions use aligned spacing on purpose, and it never joins two lines in any case.
No. It is transformed in the page.
Line endings are normalised first, so every rule after it sees plain newlines rather than the carriage returns a Windows editor leaves behind — otherwise a blank-line rule silently misses half its matches. Invisible characters go next: soft hyphens, the zero-width space and joiner family, the bidirectional override marks and the byte-order mark. These are not whitespace, so they survive a trim, and they are what makes a description look clean while breaking a search for a phrase inside it or producing a stray character when someone copies your link. Typographic substitutions straighten curly quotes, em and en dashes, the ellipsis character and the non-breaking space, all of which arrive from a word processor and none of which YouTube needs. Tracking parameters are stripped from URLs with the browser's own URL parser rather than a regular expression, so the rest of the link survives byte for byte: the query is rebuilt from the remaining parameters instead of being re-serialised, because a full re-encode would rewrite a perfectly good link. Trailing punctuation is detached before parsing, so the full stop ending the sentence after a link is not swallowed as part of it. The parameter that matters most is YouTube's own: the share button appends `si=` to every youtu.be link and it identifies the account that shared it, so a description pasted from a share sheet leaks that to everyone who reads it. Trailing spaces are then trimmed per line — a line containing only spaces is not blank to a regular expression but is blank on screen, so trimming has to come before the blank-line rule or runs of them survive. Finally three or more consecutive newlines collapse to two, which preserves the single blank line that is a real paragraph break while removing the accidental gaps. Optional space collapsing uses a character class that excludes newlines, so it can never join two lines into one. The character count is by code point against the 5,000-character limit. Everything runs in the page.