Privacymask — Data De-identification and Privacy Protection Assistant for Openclaw

1. Install or enable Privacymask inside your Openclaw Skills environment or MCP Toolbox bundle.
2. Confirm the skill is available locally; no external API or cloud connection is required.
3. Point the skill to a file or directory containing supported formats.
4. Optionally define custom masking rules before running batch jobs.
5. Choose whether to generate a masked copy, run preview-only detection, or export a report.

# Mask a single file into a new output directory
请脱敏 C:/reports/audit_2024.xlsx 输出到 C:/masked/

# Mask an entire folder
批量脱敏 C:/finance/2024/ 全部脱敏输出到 C:/finance_masked/

# Preview sensitive data without modifying the source
帮我检测 C:/data/customer_list.csv 里有哪些敏感信息

# List available masking rules
查看所有脱敏规则
查看手机号码和身份证号规则

# Add a custom rule for contract numbers
新增自定义规则：合同编号，模式为 [Tt]\d{6,}，掩码用 * 替换

# Export detection results
将检测结果导出到 C:/report.json

6. Review the generated output and exported report to verify compliance coverage before distribution.

Privacymask organizes data around three main layers: input sources, masking rules, and output artifacts.

Supported file handling behavior:

• Plain text formats: direct redaction of sensitive strings.
• CSV/TSV: cell-level masking while preserving structure.
• JSON/YAML: recursive masking of string values.
• DOCX/PPTX: paragraph, table, and slide-text masking with layout preservation.
• XLSX/XLS: cell-level masking while preserving formulas and formatting.
• PDF: text-layer redaction.

Metadata captured in reports typically includes:

• Sensitive type detected
• Count of matches
• Match position or location
• Applied rule name
• Output destination

Openclaw Skills users can extend the taxonomy with custom patterns for internal identifiers such as contract IDs, employee IDs, or proprietary account numbers.