Convert a PDF to clean markdown: headings, paragraphs, and bullets reconstructed from the text layer — ready to drop into a model's context. Body: {"url":"https://…/file.pdf"}.