documentation

PlainBytes Redactor User Guide

Written for the current Windows desktop app. If anything here disagrees with the build you have installed, trust the in-app UI.

Quick Start

The video below walks through the complete workflow from start to finish: automatic detection, with manual supplement where needed.

Full workflow demo · 2 min

1What this app is for

1.1 True redaction vs. draw-over preview

These two approaches mean very different things for compliance and security:

AspectDraw-over preview (the usual “black bar”)True redaction (what this app aims for)
Inside the PDF Text, vectors, or images can still be in the file—often just covered by another graphic. Matching content is removed from the content stream so it isn’t recoverable through normal parsing.
Copy / search Depending on the tool, “hidden” text might still copy out. The goal is that selectable text no longer contains the sensitive strings (verify on your exported file).
Best for On-screen reading or quick cover-ups. Sharing externally, archiving, audits, diligence—anywhere you need content gone for good.

Key principle

PlainBytes Redactor is built around true redaction. Use Confirm so you're clear which areas are actually removed at export.

1.2 What we mean by “local and offline”

1.3 When it’s a good fit

2Core ideas

2.1 Redaction marks (RedactionMark)

Internally, each area you might redact is a redaction mark—think “a rectangle on a page plus metadata”:

2.2 What “confirm” means—and why it’s manual

2.3 The batch baseline

3End-to-end workflow

Walkthrough setup

In this example, a legal reviewer has an 8-page employee offboarding PDF to release externally. They need mobile numbers, partial bank-card snippets, and personal email addresses removed. The file is a real electronic PDF with selectable text (not a flat scan).

Single-file redaction flow: open document, configure rules and run detection, review and add manual boxes, export, and optional batch apply.
Workflow at a glance
AOpen the document

Drag, drop, or open

Use Open or drag-and-drop to load the PDF (common image formats are also supported; the path through the app may differ). Everything—preview, detection, and marks—is tied to this one source file, so open the version you actually intend to redact.

BRegion & scenario

Pick region and scenario in the rules hub

Open Rules, choose a region (country/area), then a business scenario (general, financial, etc.). Turn built-in rules on or off, or add custom ones if needed. The effective rule set changes with region and scenario; picking the wrong scenario can leave useful rules off or invite extra noise. You can save preferences before closing so the next similar doc is quicker.

CRun detection

Let the document model run

On the right, under scan policy, click Start detection (or the equivalent control) and let the document model and full pass finish. That turns rule hits into a list of pending marks. If you change region, scenario, or rules, run detection again so the list matches.

DReview audit list

Confirm what must go

In the audit trail on the right, walk items one by one or use filters. Confirm what must go; clear confirmation or remove rows for false positives (exact controls depend on the build). The yellow overlays in preview usually stay in sync—clicking can toggle confirm state. This step is what keeps you from deleting the wrong thing or leaving something behind; export only honors confirmed marks.

EManual boxes

Catch anything rules missed

Drag rectangles on the left-hand preview over anything rules didn’t catch but you know must be redacted. Adjust selection colors if it helps on busy backgrounds. Auto detection won’t catch everything; manual boxes are your safety net.

FExport

Get your redacted copy

Choose Export redacted, pick a save location, and read the completion summary (how many items removed, metadata cleanup, validation messages). You get a share-ready copy; the summary is a quick sanity check—if it flags warnings, open the exported PDF and spot-check.

GApply to many (optional)

Reuse confirmed positions on similar files

After everything that should be redacted is confirmed on this PDF, use Apply to multiple files (or the equivalent entry). The app builds the normalized baseline described above and opens batch redaction. Add more PDFs with matching layout, set an output folder, choose whether to strip metadata, and run. When layouts truly match, you avoid redoing the same confirmations file by file; batch still physically removes the boxed regions.

Important

This isn’t a reusable named template on disk—there’s no separate template file; the baseline comes from this confirmed document. Batch also doesn’t re-run rule detection per queued file. If a file’s layout changed, go back through the single-file flow.

4Main window tour

PlainBytes Redactor main window with document preview on the left and audit and policy panels on the right; layout may vary by version.
Main window

4.1 Preview (left)

4.2 Scan policy panel

4.3 Selection appearance

4.4 Audit trail list

4.5 Toolbar (top to bottom)

The status bar shows readiness and the current file name, among other hints.

5Rules hub

5.1 Region and business scenario

5.2 Built-in vs. custom rules

5.3 When to adjust rules

6Audit sidebar

6.1 Why confirmation works this way

6.2 Filters and bulk confirm

6.3 Risk levels

Levels (e.g. critical, high, suspicious, medium, low) help you sort and filter so the riskiest hits surface first. They are not a mandate to confirm—you still decide what to redact.

7Manual selections

7.1 When to draw your own boxes

7.2 Working with auto-detected marks

A practical order: run detection → use the audit list to clear obvious false positives and notice patterns → add a few manual boxes for real misses → skim the preview page by page. Manual and automatic marks are treated the same at export time—what matters is whether they’re confirmed.

7.3 Keyboard shortcuts (preview & marks)

While the single-file page is loaded, it hooks the main window’s PreviewKeyDown. The shortcuts below apply only with a document open and the app not busy. They are not wired to the toolbar previous/next page buttons.

KeyAction
TabNext redaction mark.
Shift+TabPrevious redaction mark.
With no modifiers: previous mark, next. With any modifier held: only nudge preview scroll by a fixed step; are not handled in that branch.
EnterNo modifiers: toggle confirm on the selected mark.
DeleteNo modifiers: remove the selected mark.
Page Up/Page DownScroll the preview stack by about one viewport height.
Home/EndJump preview scroll to top / bottom.

8Export

8.1 Pre-export checklist

8.2 Why strip metadata

Beyond page content, PDFs carry author, producer, timestamps, XMP, and more. Redacting body text but leaving metadata can still leak internal accounts or edit history. This app can scrub that on export (and optionally in batch). For anything leaving the building, it’s usually best to leave metadata cleanup on when the UI offers it.

8.3 Reading the summary

9Batch processing

9.1 What batch mode actually does

After you enter through Apply to multiple files with a baseline, each queued PDF follows a stamp-only path: normalized rectangles are mapped to absolute coordinates per page, marks are created, content is physically removed, and metadata can optionally be cleared. It does not run rule detection again per file.

Tip

The rules you used in single-file mode helped you find and confirm on the template; batch simply copies where you confirmed onto other files that share the same geometry.

9.2 Good use cases

9.3 Things to watch

10FAQ

Q.Why are there so many false positives?

Rule matching leans on patterns and heuristics: strings that look sensitive, place names colliding with person names, table fragments, and so on. Tuning scenario/rules, turning off a noisy rule, or clearing confirmation in the list usually gets it under control. Manual confirmation is intentional—it lowers the odds of one-click deleting something you still need.

Q.What about scanned PDFs?

If the PDF is effectively a page image with no real text layer, rule detection has less to work with, so you’ll lean more on manual boxes or whatever raster-specific flow the build exposes (follow on-screen hints). It helps to know whether you’re dealing with a born-digital PDF or a scan up front.

Q.Can I redact images? Why isn’t text in photos recognized?

Yes—common image formats can be opened like a document; you still preview, mark, and export. To keep installer size and runtime cost reasonable, this version doesn’t ship OCR: text in photos doesn’t become selectable, so rule-based detection barely applies. Redact sensitive areas by drawing boxes (and using keyboard shortcuts to move between marks). If you need “recognize then match,” OCR elsewhere first or supply a text-layer PDF.

Q.How do I sanity-check an export?

In any PDF reader, try select all / copy and search for strings you know were sensitive. Visually scan headers, footers, and footnotes. Read the app’s export validation summary; if it warns, revisit those pages first. For very high stakes, layer your org’s usual controls—second tool, print review, etc.