본문으로 건너뛰기
cccskills
무료GitHub에 공개

data-labeling-system

Implements data classification labels and tagging systems including metadata tagging, DLP integration, automated label propagation, user-applied labels, and label inheritance rules. Covers Microsoft Purview sensitivity labels and enterprise labeling architecture. Keywords: data labeling, sensitivity labels, metadata tagging, DLP integration, label propagation, Purview, classification.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(5)

  • SKILL.md12.1 KB
  • assets/template.md3.1 KB
  • references/standards.md2.3 KB
  • references/workflows.md2.6 KB
  • scripts/process.py10.1 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

Data Classification Labels and Tagging System

Overview

Data classification labels are the operational mechanism through which classification policy is enforced across the enterprise. Labels attach classification metadata to data assets — documents, emails, database records, and cloud resources — enabling automated enforcement of handling requirements through DLP policies, access controls, and encryption. This skill covers the design and implementation of a labelling system using Microsoft Purview Information Protection as the primary platform, with architecture patterns for automated labelling, user-applied labelling, label inheritance, and cross-platform propagation.

Label Taxonomy

Vanguard Financial Services Label Hierarchy

Vanguard Classification Labels
├── Public
│   └── (no sub-labels)
├── Internal
│   └── Internal - Project Confidential
├── Confidential
│   ├── Confidential - Customer Data
│   ├── Confidential - Employee Data
│   ├── Confidential - Financial Data
│   └── Confidential - Legal
└── Restricted
    ├── Restricted - Special Category (Art. 9)
    ├── Restricted - Criminal Data (Art. 10)
    ├── Restricted - AML Investigation
    └── Restricted - Board & Strategy

Label Properties

LabelColourVisual MarkingEncryptionDLP PolicyAuto-Apply
PublicGreenFooter: "Vanguard Financial Services — Public"NoneNoneNo
InternalBlueFooter: "Vanguard Financial Services — Internal Use Only"OptionalWarn on externalNo
ConfidentialAmberHeader + Footer: "CONFIDENTIAL"Azure RMS (AES-256)Warn + audit external; block personal emailYes (when PII detected with >85% confidence)
RestrictedRedHeader + Footer: "RESTRICTED" with red background; watermark on printAzure RMS (AES-256, double key encryption)Block all external; block USB/print; alert DPOYes (when Art. 9/Art. 10 data detected with >85% confidence)

Microsoft Purview Implementation Architecture

Component Architecture

┌─────────────────────────────────────────────────────────────────┐
│                    Microsoft Purview                             │
│                                                                  │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────────┐  │
│  │ Sensitivity   │  │ Auto-labelling│  │ DLP Policies          │  │
│  │ Labels        │  │ Policies      │  │ (endpoint, email,     │  │
│  │ (definitions) │  │ (rules)       │  │  SharePoint, Teams)   │  │
│  └──────┬───────┘  └──────┬───────┘  └──────────┬───────────┘  │
│         │                  │                      │              │
│         ▼                  ▼                      ▼              │
│  ┌──────────────────────────────────────────────────────────┐   │
│  │              Unified Label Application Engine             │   │
│  └──────────────────────────────────────────────────────────┘   │
│         │                  │                      │              │
└─────────┼──────────────────┼──────────────────────┼──────────────┘
          ▼                  ▼                      ▼
   ┌─────────────┐  ┌──────────────┐  ┌──────────────────────┐
   │ Office Apps  │  │ SharePoint/  │  │ Endpoint DLP         │
   │ (Word, Excel,│  │ OneDrive     │  │ (Windows devices)    │
   │  Outlook)    │  │              │  │                      │
   └─────────────┘  └──────────────┘  └──────────────────────┘

Auto-Labelling Configuration

Policy 1: Confidential — Customer PII Detection

SettingValue
Policy NameVFS-AutoLabel-Confidential-CustomerPII
ScopeAll SharePoint sites, OneDrive accounts, Exchange mailboxes
ConditionsContent contains ANY of: UK National Insurance Number (HIGH confidence), IBAN (HIGH), Vanguard Account Number (HIGH), Credit Card (HIGH + Luhn validated)
Minimum count1 instance of any SIT at HIGH confidence
Label appliedConfidential - Customer Data
Priority2 (overridden by Restricted auto-label)

Policy 2: Restricted — Special Category Detection

SettingValue
Policy NameVFS-AutoLabel-Restricted-SpecialCategory
ScopeAll SharePoint sites, OneDrive accounts, Exchange mailboxes
ConditionsContent contains ANY of: ICD-10 codes (MEDIUM+), health terminology trainable classifier (HIGH), biometric template format detection, genetic marker patterns
Minimum count1 instance at MEDIUM confidence or above
Label appliedRestricted - Special Category (Art. 9)
Priority1 (highest priority — overrides all other auto-labels)

Policy 3: Restricted — Criminal Data Detection

SettingValue
Policy NameVFS-AutoLabel-Restricted-CriminalData
ScopeHR SharePoint sites, Compliance SharePoint sites
ConditionsContent contains: DBS reference patterns, criminal conviction terminology, SAR reference numbers
Label appliedRestricted - Criminal Data (Art. 10)
Priority1

Label Inheritance Rules

RuleDescriptionImplementation
Container inheritanceItems in a labelled SharePoint site inherit the site's label as minimumSite sensitivity label propagates to new items; existing items retain higher label
Email attachment inheritanceAttachments inherit the email's label if attachment label is lowerOutlook plugin checks attachment label vs email label on send
Parent-child inheritanceChild documents inherit parent folder label as minimumSharePoint library policy; items cannot be labelled below folder label
No downgrade without approvalUsers cannot remove or downgrade labels without justificationLabel policy: require justification text for downgrade; audit log entry
Highest label winsWhen documents are merged or combined, the highest label appliesUser training + DLP monitoring for combined documents

User-Applied Labelling

Labelling Responsibilities

ScenarioWho LabelsHow
New document creationAuthorSelect label from Office ribbon (Word, Excel, PowerPoint)
New email compositionSenderSelect label from Outlook toolbar; mandatory before sending external
File upload to SharePointUploaderLabel prompt on upload if no label detected
Data export from systemExporterLabel selection required before export completes
Physical document printingPrinterClassification header/footer printed automatically; user selects tier if not auto-labelled

Mandatory Labelling Policy

SettingValue
Require label on documentsYes — all Word, Excel, PowerPoint documents must have a label before save
Require label on emailsYes — for emails to external recipients; recommended for internal
Default labelInternal (applied if user does not select; user can override up or down)
Justification for downgradeRequired — user must enter text justification; logged in audit
Justification for removalRequired — DPO-approved exception only

DLP Integration

DLP Policy Matrix

LabelExternal EmailUSB/RemovablePrintScreenshotCloud Upload
PublicAllowAllowAllowAllowAllow
InternalWarnWarnAllowAllowBlock (non-approved cloud)
ConfidentialWarn + auditBlockSecure printAllow (watermarked)Block
RestrictedBlockBlockBlock (unless DPO approved)BlockBlock

DLP Alert Routing

Alert SeverityLabel TriggerRouting
LowInternal label — external email warn overriddenSecurity team email
MediumConfidential label — external sharing attemptedSecurity team + Data Owner
HighRestricted label — any policy triggerDPO + CISO + immediate investigation
CriticalRestricted label — data exfiltration indicatorsDPO + CISO + Incident Response Team + 15-minute SLA

Label Propagation Across Platforms

Supported Platforms

PlatformLabel MethodPropagation
Microsoft 365 (Word, Excel, PowerPoint)Native sensitivity label in file metadataFull support — label travels with file
Outlook / ExchangeEmail header X-MS-Exchange-Organization-ClassificationLabel persists in message store and forwarded copies
SharePoint OnlineDocument library metadata + file metadataDual storage — library and file level
OneDrive for BusinessFile metadataSame as SharePoint
TeamsChannel/chat message metadataLimited — file attachments inherit, message labels in compliance
PDF exportVisual marking (header/footer) + XMP metadataVisual marking persists; metadata depends on PDF viewer
Azure SQL DatabaseColumn-level sensitivity classification (sys.sensitivity_classifications)Native Azure SQL feature; integrates with Purview
AWS S3Object tags (key: classification, value: tier)AWS-native tagging; read by Macie and IAM policies
On-premises file sharesFile metadata (NTFS ADS) via AIP Unified Labelling clientRequires AIP client installed on endpoints

Enforcement Precedents

  • ICO v Interserve Group (2022): GBP 4.4 million — failure to implement adequate data classification and labelling contributed to staff not recognising the sensitivity of data compromised in a breach
  • CNIL v Free Mobile (2022): EUR 300,000 — customer personal data stored without classification or access controls; staff could access all customer data regardless of role or need

Integration Points

  • classification-policy: Labelling system is the technical implementation of the classification policy
  • auto-data-discovery: Discovery results trigger auto-labelling for newly detected PII
  • pii-in-unstructured: PII detection in documents drives label recommendation or auto-application
  • data-inventory-mapping: Labels feed the data inventory with current classification status per asset
  • data-lineage-tracking: Labels propagate through data lineage — transformed data inherits source label

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Implements 42 CFR Part 2 protections for substance use disorder patient records. Covers written consent requirements stricter than HIPAA, re-disclosure prohibition, court order procedures, qualified service organization agreements, and 2024 amendments aligning Part 2 with HIPAA. Keywords: 42 CFR Part 2, substance use disorder, SUD records, re-disclosure, consent, Part 2 amendments.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

Guides assessment of third-country adequacy decisions under GDPR Article 45 for international data transfers. Covers the current EC adequacy decisions list, adequacy assessment criteria, partial adequacy handling, and monitoring of adequacy decision reviews. Keywords: adequacy decision, Article 45, third country, adequate protection, EC adequacy list.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

Implements age-gating mechanisms for online services to restrict access based on user age. Covers hard gates versus soft gates, neutral age prompts, re-verification triggers, circumvention prevention, and regulatory requirements under GDPR, COPPA, UK Online Safety Act, and DSA. Keywords: age gate, age restriction, neutral prompt, children, online services, access control.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

Evaluates and implements age estimation and verification technologies for online services. Covers facial age estimation, digital ID verification, self-declaration with risk assessment, AI-based age estimation, and the accuracy versus privacy tradeoff. Includes ICO guidance and euCONSENT framework. Keywords: age verification, age estimation, facial analysis, digital ID, children, online safety.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

Preparing EU AI Act compliance documentation for high-risk AI systems. Covers Annex III classification, technical documentation under Art. 11, conformity assessment, risk management systems, and CE marking requirements. Keywords: EU AI Act, high-risk AI, Annex III, conformity assessment, CE marking.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

Implements GDPR Art. 22 automated decision-making and AI Act Art. 14 human oversight requirements for AI systems. Covers identification of solely automated decisions, meaningful human intervention design, logic explanation mechanisms, and contestation procedures. Keywords: Art. 22, automated decision, human oversight, AI Act, profiling, contestation.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

mukul975/Privacy-Data-Protection-Skills3142026년 3월 17일 업데이트

mukul975의 스킬 모두 보기

이 스킬의 문제 신고