AgentSkillsCN

apryse-redact

利用 Apryse SDK 永久移除 PDF 中的敏感内容。通过关键词或正则表达式模式（如社保号码、电子邮件、电话号码）进行遮盖。当用户需要对文档进行脱敏处理，或清除机密信息时，可使用此技能。

SKILL.md
--- frontmatter
name: apryse-redact
description: Permanently remove sensitive content from PDFs using Apryse SDK. Redact by search term or regex pattern (SSN, email, phone). Use when user needs to sanitize documents or remove confidential information.
license: Proprietary. LICENSE.txt has complete terms
compatibility: Requires Node.js 14+ and @pdftron/pdfnet-node. Set PDFTRON_LICENSE_KEY environment variable.
metadata:
  author: apryse
  version: "1.0"

PDF Redaction

Permanently remove sensitive content from PDFs. All scripts are in the scripts/ directory.

WARNING: Redaction is permanent and irreversible. The original content is completely removed from the file.

Available Scripts

Redact by Search Term

bash
node scripts/redact-search.js <input.pdf> <output.pdf> "term1" ["term2"] [--overlay "[REDACTED]"]

Finds and redacts all occurrences of the specified terms.

Redact by Pattern (Regex)

bash
node scripts/redact-pattern.js <input.pdf> <output.pdf> "<pattern>" [--overlay "[REDACTED]"]

Common Patterns

Data TypePatternExample Match
SSN\d{3}-\d{2}-\d{4}123-45-6789
Email[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}user@example.com
Phone (US)\(\d{3}\)\s*\d{3}-\d{4}(555) 123-4567
Credit Card\d{4}[- ]?\d{4}[- ]?\d{4}[- ]?\d{4}1234-5678-9012-3456
Date\d{1,2}/\d{1,2}/\d{2,4}12/31/2024

Examples

Redact all SSN numbers:

bash
node scripts/redact-pattern.js doc.pdf redacted.pdf "\d{3}-\d{2}-\d{4}"

Redact specific words:

bash
node scripts/redact-search.js doc.pdf redacted.pdf "confidential" "secret" "password"

Redact emails with replacement text:

bash
node scripts/redact-pattern.js doc.pdf redacted.pdf "[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}" --overlay "[EMAIL REMOVED]"

When to Use

  • •User asks to "redact", "remove", or "black out" specific text → redact-search.js
  • •User asks to "remove all SSN/email/phone numbers" → redact-pattern.js with appropriate pattern
  • •User asks to "sanitize" or "clean sensitive data" → use patterns for common PII

Notes

  • •Redaction is PERMANENT - always work on a copy
  • •The --overlay text appears in place of redacted content
  • •Patterns use JavaScript regex syntax (escape backslashes in shell)
  • •Test patterns on a copy first to verify matches