document.convert

Convert a PDF to a plain-text or markdown artifact by reading its embedded text layer. One node, one field — digital PDFs only.

Config

Field	Type	Default	Notes
`format`	`txt` · `md`	`txt`	Output packaging. `txt` is `text/plain`; `md` is `text/markdown` with a heading per page.

What it reads

PDF files that already contain a text layer. There is no OCR pass — a scan without embedded text comes back as an empty (or near-empty) file rather than failing the job. Office formats aren't accepted yet; the SDK refuses them before upload.

The result is a normal job artifact with a signed URL, the same shape as an image or audio convert — not inline JSON. Pull the bytes from media.url within the expiry window.

In a pipeline

{
  id: 'convert',
  type: 'document.convert',
  config: {
    format: { value: 'md' },
  },
}

With runQueue

tk.runQueue(files, 'document')
  .convert({ format: 'md' });

Every output is a job and a credit, the same as pipeline.output describes.