Skip to main content

MIME Filter

Automatically sorts each support document by its type — deterministic detection by extension/content, no AI call involved. Never applies to the main document, only to support documents.

Ports​

  • Input (left, support document type).
  • 3 fixed outputs (not configurable):
    • Image — requires a downstream OCR node.
    • Scanned PDF — also requires OCR.
    • Other — already usable (native PDF, Word, Excel…), goes directly back to the "support documents" input of the Document to Annotate node.

Configuration​

None.

Refusing a category​

Each output can also lead to an Error message wired to the Error exit: a support document in that category gets the job rejected with your message, before any other step (OCR does not run). For instance, Image → "Send the ID document as a PDF, not a photo" refuses photos while letting PDFs through. Images are then accepted at upload even without an OCR node, so that the user gets your message rather than a generic refusal.

In chat, only the Image output applies, as soon as the attachment is sent — see Error message.