[BrianGenisio]

Matching suggested answers with $$...$$ now render as math on the choice chips and on the selected card. Scoring still compares the raw suggested-answer string to [[blank:...]].

Changes

Matching choices used textContent, so learners saw dollar signs instead of formulas. The parser now includes choicesHtml for each chip. choices stays the raw string so blank matching is unchanged.

The chip is a flex .button. Chrome treated the sentence and the KaTeX tree as sibling flex items, which wrapped into two columns and split y=x^2. Content now sits in a single .matching-choice-label block, and the formula is white-space: nowrap.

Testing

• npm test
• npm run examples, open matching-latex-chips.md
• Confirm suggested answers render as math, not $$y=x^2$$
• Place a chip on a card. When the label wraps, the formula should stay on one line with the surrounding sentence, not in a second column

[BrianGenisio]

@coderrabbit review