I think your approach is reasonable. But I'm curious, why provide the challenge in PDF rather than HTML? Unless your PDF generator can create a tagged PDF, that

I think your approach is reasonable. But I'm curious, why provide the challenge in PDF rather than HTML? Unless your PDF generator can create a tagged PDF, that's going to be less accessible for blind people to blind people than HTML, .docx or .odt, or even plain text. I don't mean to guilt-trip here; I figure you weren't aware of that problem. Is there a reason why PDF is better? For example, do your challenges tend to include graphics or complex math notation?