PDF Redactor
Availability: not available. At the time of writing, PDF Redactor has no web page, no CLI plugin, no MCP tool and no API endpoint in Documentize. The product's own sources show an empty project folder for it, and the capability catalogue lists "redact" as a known gap. This chapter explains what redaction is, what to do today, and how to check that your workaround really worked. If the app is released, this chapter will gain the usual step-by-step; check the live site and What's new.
In one sentence
Redaction means permanently removing sensitive text or images from a document (not just covering them), and Documentize cannot do it for you yet.
What makes it unique
Nothing yet: there is no tool to compare. What matters is the concept, because most people redact wrongly:
- Covering is not redacting. A black rectangle drawn over a name leaves the name underneath; anyone can select it, copy it or remove the rectangle.
- Real redaction deletes the content from the file, then (optionally) paints a black box in its place.
- A password (Protector) is not redaction either: whoever has the password sees everything.
When to use it / when not to
- You need redaction when you must share a document that contains names, numbers, addresses, health or financial data that the reader should never see.
- You do not need it when whole pages are the problem. Cut them out with Delete Page or keep only the pages you want with Splitter. That is permanent, because those pages are not in the result.
- You do not need it when the whole file is only for chosen people. Use Protector and share the password separately.
Before you start
- Decide what must go: text, images, metadata (author names, comments), whole pages.
- Keep the unredacted original somewhere safe, and send only the redacted copy.
- If the document is a scan, the sensitive words are pixels, not text; a redaction tool would need to black out regions, and any text layer added earlier by Searchable must be removed too.
Step by step
What you can do today, using Documentize for the parts it does well:
- Remove whole pages with sensitive content: Delete Page or Splitter.
- Check metadata with Extract Metadata; author, title and custom properties can reveal names.
- For text inside a page, use a dedicated redaction tool outside Documentize (a desktop PDF editor with true "apply redactions"). Make sure it removes the underlying text, not only paints over it.
- Verify with Extract Text or Search: the redacted words must not be found.
- Flatten with Flatten to bake in remaining annotations or form fields, then optionally protect with Protector.
📷 Screenshot placeholder: none, because there is no page.
Options explained
| Option | What it does | Default | Recommended |
|---|---|---|---|
| – | No options exist, because the app does not exist | – | – |
What you get
Nothing from Documentize for this step. The outcome you want is a new PDF in which the sensitive content cannot be recovered; confirm it with the verification step above.
Quality & limits
- Without a verified redaction tool, treat any "redacted" PDF as unsafe until you have proved, with a text search and a look at the metadata, that the sensitive items are gone.
- Redaction is irreversible by design; always keep the original separate.
- AI summarisers are not redactors; never rely on them to hide data.
Works well with
- Use before: Delete Page, Splitter, Extract Metadata.
- Use after: Flatten, Protector, Sign.
- Recipes: Share only part of a document, Safe external sharing checklist, Contract round-trip (these recipes name Redactor as a step; until it exists, replace that step with the manual route above).
Common mistakes & fixes
- "I drew black boxes with an annotation tool." The text is still there. Test with a text search; if found, it is not redacted.
- "I protected the file with a password instead." That limits who can read it, not what they see.
- "I forgot the metadata and comments." Check them before sending.
- "I sent the original by accident." Name files clearly, such as
contract_REDACTED.pdf.
Same thing elsewhere
- CLI: none. MCP: none. API: none.
- The platform's own advisor lists redaction as a named gap rather than pretending otherwise.
Try it
Make a one-page PDF that contains a fake phone number. Search for the number with Search so you know it is found. Now "hide" it by any method you have (a drawn rectangle or a covering shape), save, and search again. If it is still found, you have just seen why covering is not redacting.