Markdown bold that ends in punctuation cannot close right before Korean, Japanese or Chinese text

Fact

In CommonMark and GFM, a closing * or ** that follows punctuation counts as closing only when whitespace or more punctuation comes after it. A Hangul particle, kana or hanzi is neither, so **"사과"**를, **「東京」**に and **apple.**은 render with the asterisks left in as literal text. With no punctuation at the edge, the same run closes: **사과**를 renders bold.

The opening side mirrors it: ** after a letter and before punctuation cannot open, and the stray delimiters can then pair with a later ** and bold a span nobody marked.

Why it matters

Korean attaches particles directly to the word, and Japanese and Chinese use no spaces at all, so the failing position is the normal one in CJK prose. Nothing errors: GitHub.com shows raw ** or bolds the wrong text, and only on the lines where the emphasized text happens to start or end with a quote, bracket or period.

How to apply

• Keep punctuation outside the emphasis: "**사과**"를, **사과**(apple)를.
• When the punctuation belongs inside, use HTML: <strong>"사과"</strong>를.
• Use *, never _, next to CJK text: underscores do not close inside a word at all, so _사과_를 fails even with no punctuation.
• Do not add a space before the particle to make it render; that changes the text.
• Check a doubtful line with GitHub's own renderer: gh api markdown -f mode=markdown -f text='**"사과"**를'.