Skip to content
Ghostchars

Bidirectional marks

LEFT-TO-RIGHT MARK (U+200E)

An invisible, zero-width character with strong left-to-right direction. It resolves ambiguity in mixed text without reordering anything itself.

Codepoint
U+200E
Removed by default
When free-floating
Governing option
Bidi marks --bidi
Can be load-bearing
Yes

Where it comes from

Bidirectional documents, where it fixes a punctuation mark that would otherwise jump to the wrong end of a sentence. Also from CMSes and translation tools that add it defensively, and from bidi-mark steganography, which encodes bits as U+200E and U+200F.

Why it matters

Next to Hebrew or Arabic it is doing real work and removing it visibly breaks the layout. In an English-only paragraph it is invisible padding — and a pair of invisible characters with two states is a one-bit-per-position channel.

Before and after

BeforehelloU+200E world
Afterhello world

No right-to-left text anywhere near it, so the mark has no ambiguity to resolve and is removed.

The character is shown as a labelled chip so you can see where it sits. In your text it draws nothing at all.

When it is legitimate

Immediately before or after right-to-left text. Ghostchars checks the neighbours for a right-to-left, Arabic-letter or Arabic-number character and keeps the mark when it finds one.

How to remove it

A mark with no right-to-left neighbour is removed by the default pass. One next to right-to-left text is kept and reported; `--bidi` removes those too, and will change how mixed Hebrew or Arabic renders.

Is it an AI watermark?

Used to hide data — not a watermark

No. LRM/RLM binary is a published steganography scheme, and the mark is also ordinary bidi punctuation. Context decides which you are looking at; authorship never enters into it.

A known carrier for deliberately hidden payloads. Someone put it there on purpose, which is a different question from who wrote the words.

Questions

Will cleaning break my Hebrew or Arabic layout?

Not by default. A mark adjacent to right-to-left text is kept and reported as load-bearing. `--bidi` is the option that removes it anyway, and it says what it breaks.

How is it used to hide data?

By placing a LRM for a 0 and a RLM for a 1 after each word. Both are invisible in a left-to-right paragraph, so the text looks untouched.

Check your own text

Paste anything into the cleaner to see every hidden character it contains, with position, codepoint and what happens to each one. Nothing is uploaded.

Check your own textAll characters