Tutorial: Redact a PDF Fetched Over HTTP

A complete, runnable n8n workflow that fetches a PDF over HTTP and redacts a face in it with the PDF Redaction node.

This tutorial builds a complete, runnable n8n workflow that:

Fetches a real PDF over HTTP with the HTTP Request node.
Redacts the faces in it with the PDF Redaction node.
Shows the redacted file and detection results in the output panel.

It uses a public sample file from the pdf-redaction-api repository — SampleWithFace.pdf — so you can reproduce every step with no files of your own.

For the general setup/configuration walkthrough, see Setup & Configuration Tutorial first. This guide assumes the node is installed and you have a PDF Redaction API credential already.

1. Trigger + HTTP Request

Start with a Manual Trigger, then add an HTTP Request node with:

Method: GET
URL: https://raw.githubusercontent.com/StabRise/pdf-redaction-api/main/examples/pdfs/SampleWithFace.pdf
Options → Response → Response Format: File

Execute the node. The response comes back as binary data in the data field:

HTTP Request node output showing the fetched SampleWithFace.pdf binary data

2. PDF Redaction — Anonymize the "Face" tag

Add a PDF Redaction node after the HTTP Request node. It picks up the data binary field automatically (both Input Binary Field and Output Binary Field default to data, matching what HTTP Request produced).

Set:

Credential: your PDF Redaction API credential
Operation: Anonymize
Additional Fields → Tags: Face

PDF Redaction node configured with the Anonymize operation and Face tag, reading input from the HTTP Request node

3. Execute and inspect the output

Click Execute step. The node calls the PDF Redaction API and returns:

A redacted PDF (redacted.pdf) in the output binary field, with the face blacked out.
A JSON payload with detected_pii (entities found, here two FACE entries with bounding boxes) and a processing_time breakdown per pipeline stage.

JSON output:

Output JSON panel showing detected_pii FACE entities with bounding boxes and processing_time for the anonymize run

Binary output:

Output Binary panel showing the redacted.pdf file, 377 kB

The original file was 823 kB; the redacted output is 377 kB, since the API flattens the processed page(s) to a compact PDF.

4. The processed PDF

Opening the downloaded redacted.pdf confirms both faces on the page are blacked out:

Rendered page of the redacted PDF, showing both faces blacked out

5. The complete workflow

All three nodes wired together, each showing a successful execution (green checkmarks):

Complete workflow: Manual Trigger → HTTP Request → PDF Redaction, all executed successfully

Adapting this to your own files

Swap the HTTP Request node for whatever produces your PDF as binary data — a webhook payload, a Read/Write File from Disk node, an email attachment (via an email trigger), or a file from cloud storage (S3, Google Drive, etc.). As long as the upstream node writes binary data to a field, and the PDF Redaction node's Input Binary Field matches that field name, this same pattern applies.