LLM Math Delimiters vs Markdown Parsers: A Regex Fails, Tokenizer Wins

LLMs often output math in TeX delimiters like \(...\) and \[...\], but most Markdown math plugins only accept $...$. A developer tried regex-based preprocessing, which broke on edge cases like TeX commands and unclosed delimiters. Instead, they built two packages that extend micromark and remark to parse the TeX-style delimiters natively.