Comments sent to LanguageTool are not stripped of their language-specific comment marks

For instance, for https://github.com/dustinblackman/languagetool-code-comments/blob/main/tests/fixtures/astro.astro#L7, the text that gets sent to the LanguageTool server for analysis is <!-- I am a comment number two. -->, where I would have expected that I am a comment number two. be sent.
This can result in LanguageTool reporting various language issues, such as a Do you wish to insert an arrow? typography issue.

Although I understand that stripping the language-specific comment marks may not be trivial (because it is language-specific), the current behaviour is surely not optimal.