I guess these days most projects use GitHub markdown since it's the most common and with the best documentation.
Of course nowadays we have CommonMark, but despite ongoing activity CM has yet to have a 1.0 release, meaning that even if one intended to adhere to CM, ongoing maintenance would be required to ensure compatibility. Given that it looks as if marked.js last had a release in mid-2015, that would lead one to assume that it's not trying to follow CM. Furthermore, it was in March of this year that GFM was officially redefined to be an extension of CommonMark, not of Markdown: https://github.github.com/gfm/ , so it's unclear which version of GFM marked.js is emulating.
Lest anyone think I'm taking this too seriously though, the project here in the OP almost certainly doesn't need to care about compatibility. The markup produced is completely transient, so the worst that would happen upon hitting an unintended edge case would just be some temporary cursing on the part of the user. Where it actually matters is when you're storing the markup long-term and expect to ever want to upgrade your Markdown parser without breaking everything; for an example of the difficulty here, see how Rust has been attempting to gracefully upgrade the parser in rustdoc for almost a year now: https://internals.rust-lang.org/t/what-to-do-about-pulldown-... , https://github.com/rust-lang/rust/issues/44229
MDH-specific cheatsheet here: https://github.com/adam-p/markdown-here/wiki/Markdown-Here-C...