Invisible Unicode trick stops X from auto-linking domain names

Flavio Copes reports that adding a zero-width non-joiner, Unicode U+200C, immediately after the last dot in a bare domain keeps X from turning it into a clickable link. He points to a Paul Graham post that Guillermo Rauch flagged on X and says the post returns no link entities from the X API. Copes ties the behavior to twitter-text, X’s open-source parser, which accepts only ASCII letters, digits, hyphens and dots for bare domains ending in a known TLD; inserting U+200C breaks that pattern. He also shows the trick is fragile: put the character in the wrong place and X may still link a different valid domain, while copied text carries the hidden character and can become invalid in browsers and other parsers.