Developer docs

Introduction

Redact PDF AI developer docs: the REST API for PDFs, scans and photos (async jobs, X-API-Key), redact or pseudonymize mode, and the MCP server for AI assistants.

Introduction

Redact PDF AI finds the personal data in PDFs, scans and photos and removes it for good: the text is deleted from the file, not covered with a box. Use it through the REST API, from an AI assistant over MCP, or in the web app.

  • Base URL: https://www.redact-pdf.ai
  • Auth: X-API-Key header, from server-side code only
  • Format: asynchronous jobs (upload, poll, download)
  • Input: PDF, JPEG and PNG, scans and phone photos included
  • Output: a flattened, redacted PDF with black boxes, or with consistent tags such as [Person 1] in pseudonymize mode
  • OpenAPI: openapi.yaml
  • Processing on Microsoft Azure in the EU (Ireland) and Switzerland; documents are never used to train AI models
A short tour: upload, check what the AI found, adjust, download.

Start here

Core flow

  1. Create a job with one or more files: POST /v1/jobs.
  2. Poll the job until each document is redacted or error: GET /v1/jobs/{job_id}.
  3. Download each redacted document: GET /v1/documents/{doc_id}/output.
  4. Optionally delete the job to purge everything now: DELETE /v1/jobs/{job_id}.

Document status: uploaded → analyzing → redacting → redacted (done) or error.

Redact or pseudonymize

  • Redact (default) draws black boxes. The removed text is gone from the file.
  • Pseudonymize (mode=pseudonymize) replaces each person, company and address with a numbered tag, the same on every page, so the document stays readable. You also get the document as text and a table that maps each tag back to the original. See the Pseudonymize guide.

API reference

Guides

AI assistants

Resources