Run the pipeline

Lint a paper.

Upload a PDF or give an arXiv id. The file goes through the real compiler and the real rules — the same code path the command line uses. Compiling takes a couple of minutes because a model reads the paper once; you will see which stage is running rather than a spinner.

The LaTeX → HTML render of the paper is fetched from arXiv — most recent papers have one, but not all. If yours does not, the run will say it could not read the paper; upload the PDF instead. Use one or the other, not both.