LocalMask Documentation
Product overview
LocalMask is built for people who need to remove or alias sensitive information before documents leave their computer. It is not a cloud PII API, a browser-only paste box, a developer framework or a simple PDF annotation layer.
The product focuses on local processing, human review, content-level redaction, exportable clean copies and local alias restoration for AI-assisted or externally shared work.
Use cases
Remove or alias direct identifiers before sending contracts, reports, tables, screenshots or copied text to AI tools.
Prepare contracts, letters, case files, evidence packages and client documents before external sharing.
Handle HR records, resumes, payroll files, invoices, reimbursement material, bank statements, medical reports and internal reports.
Prepare health check reports, medical summaries, lab reports and related attachments before external sharing or AI-assisted review.
Review tenders, bids, procurement files, supplier documents, quotation tables and evaluation material before sharing, archiving or AI-assisted analysis.
Review Chinese, English or mixed-language material before sharing it with outside counsel, vendors or cloud systems.
Use one desktop workflow for PDF, Word, Excel, CSV, scans, images and clipboard text instead of separate tools for each format.
Core capabilities
Document import, detection, review, redaction and export run on the user's computer.
Detected entities, categories, locations and decisions can be inspected before a clean copy is exported.
LocalMask focuses on processing text layers and image pixels, not only drawing a visible overlay.
Names, organizations, addresses and accounts can be replaced with stable aliases so the document remains useful for analysis.
AI output or supported files containing aliases can be restored from the local mapping on the same computer.
Copied text can be detected locally and written back in redacted form before pasting into AI tools or web forms.
Supported materials
The examples below are not an exhaustive list. Any supported PDF, Word, Excel, CSV, TXT, Markdown, image, scan or clipboard text can be imported and reviewed; detection quality depends on file quality, layout complexity and the information present in the material.
Text PDFs, scanned PDFs, image-based PDFs, case files, reports, tender documents, bid documents and evidence material.
Contracts, letters, memos, tenders, bids, procurement files, statements, reports and review drafts.
Spreadsheets, customer lists, supplier lists, quotation tables, procurement ledgers, HR data, finance tables and operational exports.
Screenshots, document photos, ID images, chat screenshots and scanned pages.
Plain text, notes, prompts, drafts and structured text files.
Website excerpts, chat or email snippets, AI prompts and web form content.
Sensitive information coverage
Detection quality depends on document quality, layout complexity, scan clarity, handwriting, context ambiguity and user review.
Names, ID numbers, travel document numbers, phone numbers, emails, URLs, social accounts, addresses, postal codes, vehicle plates, VINs and IP addresses.
Company names, organization names, business identifiers, DUNS numbers, patent agency codes, case numbers and document reference numbers.
Bank accounts, card numbers, stock codes, invoice numbers, property certificate numbers, amounts and dates.
Seals, fingerprints, faces, QR codes and barcodes.
User-defined blacklists and whitelists for business-specific terms, public names and terms that should not be processed.
AI preparation workflow
- 01
Import a document or copy the text that will be sent to an AI tool.
- 02
Detect sensitive entities locally and review the results.
- 03
Replace identifiers with aliases such as Person A, Company A or Address A when context should be preserved.
- 04
Export a redacted copy or paste redacted clipboard text into the AI tool.
- 05
When AI output contains aliases, restore the aliases locally from the saved mapping if needed.
Clipboard redaction
Sensitive data often leaks through copied text, not only through file uploads. LocalMask can inspect copied text locally and write a redacted version back to the system clipboard before the user pastes it into an AI tool, web form, email thread or collaboration system.
- Copied website, chat, email or document snippets are detected locally.
- Stable aliases can preserve useful context while hiding raw identifiers.
- The redacted version can be written back before pasting into external tools.
- When needed, the user can continue review in the LocalMask workspace.
Platform, installation and licensing
System requirements
Windows 10 64-bit or later is recommended.
8GB or more memory is recommended for large PDFs and scans.
macOS 12 Monterey or later is recommended.
Apple Silicon and Intel builds are available.
Linux x64 and Linux ARM64 builds are available for Chinese and enterprise scenarios.
Target distributions should be validated before formal rollout.
No dedicated GPU or CUDA is required.
Integrated graphics are sufficient.
Comparison with PDF editors
Ordinary PDF editors focus on manual annotation or editing a single file.
LocalMask is designed for detection, review, redaction, export and alias restoration.
Visual boxes may leave hidden text searchable, selectable or extractable depending on the file and tool.
LocalMask focuses on content-level processing for exported copies.
Multiple files and scanned pages often require manual inspection one by one.
PDF, Word, Excel, CSV, images, scans and text are handled in one workflow.
Black boxes can remove useful context from the document.
Aliases preserve relationships while hiding direct identifiers.
Decisions are scattered across the document.
Detected entities and decisions are reviewed in a centralized workspace.
Privacy and security boundaries
- Original files, detection results, review decisions and alias mappings stay in the local workspace.
- LocalMask is intended to prepare sensitive material before it is sent to AI tools, cloud services, web forms or external collaborators.
- Users should review results before final delivery, especially for low-quality scans, complex layouts, handwriting and domain-specific names.
- LocalMask does not bypass unknown passwords, does not crack protected files and does not replace the user's compliance judgment.
Frequently asked questions
Does LocalMask need an internet connection?
No. Document import, detection, review, redaction, export, alias restoration and activation run on the local computer. Internet access is only needed to download the installer or purchase a license.
Will my original files be uploaded?
No. Original files, detection results, review decisions, processing records and alias mappings stay in the local workspace.
Is LocalMask suitable before using ChatGPT or other AI tools?
Yes. LocalMask is designed to remove or alias sensitive information locally before content is sent to AI tools, cloud services, web forms or external collaborators.
Is the redaction only a visual overlay?
No. LocalMask focuses on content-level redaction for exported copies, processing text layers and image pixels instead of only drawing a visible box.
Does it support English documents?
Yes. LocalMask supports common sensitive information in English and Chinese material, including names, emails, phones, URLs, addresses, organizations, business identifiers, accounts, amounts and dates.
Can LocalMask handle tender or procurement documents?
Yes. LocalMask can process PDFs, Word files, Excel files, CSV files, scans and images used for tenders, bids, procurement files, supplier documents and quotation tables. Some procurement or bidding information may be intended for disclosure, so users should review and confirm the final redaction scope before export.
Is manual review still needed?
Yes. Users should review matches before export, especially for low-quality scans, complex layouts, handwriting and domain-specific names.