blind

Replace sensitive data in documents with consistent fictional pseudonyms while preserving structure.

1|Updated Jun 29, 2026
One-click install
npx skills add https://github.com/natthasath/natthasath-marketplace --skill blind-natthasath
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: blind
Source: https://github.com/natthasath/natthasath-marketplace/tree/main/plugins/document/skills/blind
Command: npx skills add https://github.com/natthasath/natthasath-marketplace --skill blind-natthasath

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Sharing documents that contain personal names, phone numbers, emails, organization names, or financial figures risks exposing sensitive information. This Skill pseudonymizes documents so they can be shared safely while remaining readable and contextually accurate. ## Core Features & Use Cases - Three Blind Levels: Light (personal identifiers), Standard (adds organization names, URLs, addresses, document numbers), and Full (adds dates, financial figures, job titles). - Consistent Mapping: Scans the entire document first so every real value maps to the same fictional value throughout, preserving relationships, hierarchy, timelines, and quantitative logic. - Multi-Format Support: Works with pasted text and attached files including xlsx, csv, docx, and pptx, returning a blinded document plus a confidential mapping table. - Use Case: You need to share a Thai government procurement document with an external consultant. Run /blind at Standard level to replace all names, agency names, and reference numbers with realistic fictional values while keeping the approval chain and budget logic intact. ## Quick Start Invoke /blind with your document or pasted text, then choose Light, Standard, or Full when prompted to receive the blinded version and a separate mapping table.

Frequently Asked Questions about blind

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I anonymize a document before sharing it externally?▼

Invoke /blind with the pasted text or an attached file, then choose a blind level. The skill scans the full document, builds a consistent mapping of real to fictional values, and returns a blinded version plus a separate confidential mapping table.

What file formats does document pseudonymization support?▼

The blind skill accepts directly pasted text and attached files in xlsx, csv, docx, and pptx formats. Multiple documents can share one mapping set if you notify the skill beforehand.

What is the difference between light, standard, and full anonymization levels?▼

Light replaces personal names, phone numbers, emails, and ID numbers. Standard adds organization names, URLs, addresses, document numbers, and project names. Full additionally shifts dates and financial figures while preserving their relative order and proportions.

Does pseudonymization keep the document structure and meaning intact?▼

Yes. The skill preserves formatting, organizational hierarchy, approval chains, chronological order, and quantitative relationships such as budget comparisons. Only sensitive values change; the document remains fully readable and contextually coherent.

Can the original data be recovered from a blinded document?▼

The blinded document alone is designed to be non-reversible since identifiers use unrelated fictional values. However, the separate mapping table allows reversal, so it must be kept confidential and never attached with the blinded document.