MCP HubMCP Hub
SKILL·B8F01D

monitor-data-integrity

pjt222
업데이트됨 1 month ago
9 조회
26
3
26
GitHub에서 보기
기타data

정보

이 스킬은 개발자가 GxP 시스템을 위한 ALCOA+ 원칙 기반 데이터 무결성 모니터링 프로그램을 설계하고 운영하는 데 도움을 줍니다. 탐지 통제, 감사 추적 검토 계획, 이상 징후 탐지 패턴 및 에스컬레이션 매트릭스를 제공합니다. 모니터링 프로그램 구축 시, 데이터 무결성에 초점을 맞춘 점검 준비 시, 또는 MHRA나 WHO와 같은 규제 가이드라인을 이행할 때 사용하세요.

빠른 설치

Claude Code

추천
기본
npx skills add pjt222/agent-almanac -a claude-code
플러그인 명령대체
/plugin add https://github.com/pjt222/agent-almanac
Git 클론대체
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/monitor-data-integrity

Claude Code에서 이 명령을 복사하여 붙여넣어 스킬을 설치하세요

문서


name: monitor-data-integrity description: > Datenintegritaets-Ueberwachungsprogramm auf Basis von ALCOA+-Prinzipien entwerfen und betreiben. Umfasst Detektivkontrollen, Auditpfad-Pruefungsplaene, Anomalieerkennungsmuster (ausserhalb der Geschaeftszeiten, sequenzielle Aenderungen, Massenanpassungen), Metrik-Dashboards, Untersuchungsausloeser und Definition der Eskalationsmatrix. Anzuwenden beim Aufbau eines Datenintegritaets-Ueberwachungsprogramms fuer GxP-Systeme, bei der Inspektionsvorbereitung wo Datenintegritaet ein Schwerpunkt ist, nach einem Datenintegritaetsvorfall mit erweiterter Ueberwachung oder bei der Implementierung von MHRA-, WHO- oder PIC/S-Leitlinien. locale: de source_locale: en source_commit: 6f65f316 translator: claude-opus-4-6 translation_date: 2026-03-16 license: MIT allowed-tools: Read Write Edit Bash Grep Glob metadata: author: Philipp Thoss version: "1.0" domain: compliance complexity: advanced language: multi tags: gxp, data-integrity, alcoa, monitoring, anomaly-detection, compliance

Datenintegritaet ueberwachen

Ein Programm entwerfen und betreiben, das Datenintegritaet ueber validierte Systeme hinweg kontinuierlich ueberwacht, unter Anwendung von ALCOA+-Prinzipien und Anomalieerkennung.

Wann verwenden

  • Aufbau eines Datenintegritaets-Ueberwachungsprogramms fuer GxP-Systeme
  • Vorbereitung einer Behoerdeninspektion, bei der Datenintegritaet ein Schwerpunktbereich ist
  • Nach einem Datenintegritaetsvorfall mit erweiterter Ueberwachung
  • Regelmaessige Ueberpruefung bestehender Datenintegritaetskontrollen
  • Implementierung der MHRA-, WHO- oder PIC/S-Datenintegritaetsleitlinien

Eingaben

  • Erforderlich: Systeme im Umfang und ihr ALCOA+-Risikoprofil
  • Erforderlich: Anwendbare Leitlinien (MHRA Data Integrity, WHO TRS 996, PIC/S PI 041)
  • Erforderlich: Aktuelle Auditpfad-Faehigkeiten jedes Systems
  • Optional: Frueherer Datenintegritaetsbefunde oder regulatorische Beobachtungen
  • Optional: Bestehende Ueberwachungsverfahren oder Metriken
  • Optional: Benutzerzugangsmatrizen und Rollendefinitionen

Vorgehensweise

Schritt 1: Aktuelle ALCOA+-Situation beurteilen

Jedes System gegen alle ALCOA+-Prinzipien bewerten:

# Data Integrity Assessment
## Document ID: DIA-[SITE]-[YYYY]-[NNN]

### ALCOA+ Assessment Matrix

| Principle | Definition | Assessment Questions | System 1 | System 2 |
|-----------|-----------|---------------------|----------|----------|
| **Attributable** | Who performed the action and when? | Are all entries linked to unique user IDs? Is the timestamp system-generated? | G/A/R | G/A/R |
| **Legible** | Can data be read and understood? | Are records readable throughout retention period? Are formats controlled? | G/A/R | G/A/R |
| **Contemporaneous** | Was data recorded at the time of the activity? | Are timestamps real-time? Are backdated entries detectable? | G/A/R | G/A/R |
| **Original** | Is this the first-captured data? | Are original records preserved? Is there a clear original vs copy distinction? | G/A/R | G/A/R |
| **Accurate** | Is the data correct and truthful? | Are calculations verified? Are transcription errors detectable? | G/A/R | G/A/R |
| **Complete** | Is all data present? | Are deletions detectable? Are all expected records present? | G/A/R | G/A/R |
| **Consistent** | Are data elements consistent across records? | Do timestamps follow logical sequence? Are versions consistent? | G/A/R | G/A/R |
| **Enduring** | Will data survive for the required retention period? | Is the storage medium reliable? Are backups verified? | G/A/R | G/A/R |
| **Available** | Can data be accessed when needed? | Are retrieval procedures documented? Are access controls appropriate? | G/A/R | G/A/R |

Rating: G = Good (controls adequate), A = Adequate (minor improvements needed), R = Remediation required

Erwartet: Jedes System hat eine bewertete ALCOA+-Bewertung mit spezifischen Erkenntnissen fuer jedes Prinzip. Bei Fehler: Kann ein System nicht bewertet werden (z. B. keine Auditpfad-Faehigkeit), es als kritische Luecke kennzeichnen, die sofortige Behebung erfordert.

Schritt 2: Detektivkontrollen entwerfen

Die Ueberwachungsaktivitaeten definieren, die Datenintegritaetsverletzungen erkennen:

# Detective Controls Design
## Document ID: DCD-[SITE]-[YYYY]-[NNN]

### Audit Trail Review Schedule
| System | Review Type | Frequency | Reviewer | Scope |
|--------|-----------|-----------|----------|-------|
| LIMS | Comprehensive | Monthly | QA | All data modifications, deletions, and access events |
| ERP | Targeted | Weekly | QA | Batch record modifications and approvals |
| R/Shiny | Comprehensive | Per analysis | Statistician | All input/output/parameter changes |

### Review Checklist
For each audit trail review cycle:
- [ ] All data modifications have documented justification
- [ ] No unexplained deletions or void entries
- [ ] Timestamps are sequential and consistent with business operations
- [ ] No off-hours activity without documented justification
- [ ] No shared account usage detected
- [ ] Failed login attempts are within normal thresholds
- [ ] No privilege escalation events outside change control

Erwartet: Detektivkontrollen sind terminiert, zugewiesen und mit klaren Pruefkriterien dokumentiert. Bei Fehler: Werden Auditpfad-Ueberpruefungen nicht termingerecht durchgefuehrt, die Luecke dokumentieren und an das QA-Management eskalieren. Versaeumte Pruefungen haeufen Risiken an.

Schritt 3: Anomalieerkennungsmuster definieren

Spezifische Muster erstellen, die Untersuchungen ausloesen:

# Anomaly Detection Patterns

### Pattern 1: Off-Hours Activity
**Trigger:** Data creation, modification, or deletion outside business hours (defined as [06:00-20:00 local time, Monday-Friday])
**Threshold:** Any GxP-critical data modification outside defined hours
**Response:** Verify with user and supervisor within 2 business days
**Exceptions:** Documented shift work, approved overtime, automated processes

### Pattern 2: Sequential Modifications
**Trigger:** Multiple modifications to the same record within a short timeframe
**Threshold:** >3 modifications to the same record within 60 minutes
**Response:** Review modification reasons; verify each change has documented justification
**Exceptions:** Initial data entry corrections within [grace period, e.g., 30 minutes]

### Pattern 3: Bulk Changes
**Trigger:** Unusually high volume of data modifications by a single user
**Threshold:** >50 modifications per user per day (baseline: [calculate from normal usage])
**Response:** Verify business justification for bulk activity
**Exceptions:** Documented batch operations, data migration activities under change control

### Pattern 4: Delete/Void Spikes
**Trigger:** Unusual number of record deletions or voidings
**Threshold:** >5 delete/void events per user per week
**Response:** Immediate QA review of deleted/voided records
**Exceptions:** None — all delete/void events require documented justification

### Pattern 5: Privilege Escalation
**Trigger:** User access changes granting administrative or elevated privileges
**Threshold:** Any privilege change outside the user access management SOP
**Response:** Verify with IT security and system owner within 24 hours
**Exceptions:** Emergency access per documented emergency access procedure

### Pattern 6: Audit Trail Gaps
**Trigger:** Missing or interrupted audit trail entries
**Threshold:** Any gap > 0 entries (audit trail should be continuous)
**Response:** Immediate investigation — potential system malfunction or tampering
**Exceptions:** None — audit trail gaps are always critical

Erwartet: Muster sind spezifisch, messbar und handlungsrelevant mit definierten Schwellenwerten und Reaktionsverfahren. Bei Fehler: Sind Schwellenwerte zu niedrig (zu viele Fehlalarme), auf Basis von Basisdaten anpassen. Sind sie zu hoch (echte Probleme werden uebersehen), nach dem ersten Ueberwachungszyklus verschaerfen.

Schritt 4: Metrik-Dashboard aufbauen

# Data Integrity Metrics Dashboard
## Document ID: DIMD-[SITE]-[YYYY]-[NNN]

### Key Performance Indicators

| KPI | Metric | Target | Yellow Threshold | Red Threshold | Source |
|-----|--------|--------|-----------------|---------------|--------|
| DI-01 | Audit trail review completion rate | 100% | <95% | <90% | Review log |
| DI-02 | Anomalies detected per month | Trending down | >10% increase MoM | >25% increase MoM | Anomaly log |
| DI-03 | Anomaly investigation closure rate | <15 business days | >15 days | >30 days | Investigation log |
| DI-04 | Open data integrity CAPAs | 0 overdue | 1-2 overdue | >2 overdue | CAPA tracker |
| DI-05 | Shared account instances detected | 0 | 1-2 | >2 | Access review |
| DI-06 | Unauthorised access attempts | <5/month | 5-10/month | >10/month | System logs |
| DI-07 | Audit trail gap events | 0 | N/A | >0 (always red) | System monitoring |

### Reporting Cadence
| Report | Frequency | Audience | Owner |
|--------|-----------|----------|-------|
| DI Metrics Summary | Monthly | QA Director, System Owners | QA Analyst |
| DI Trend Report | Quarterly | Quality Council | QA Manager |
| DI Annual Review | Annual | Site Director | QA Director |

Erwartet: Dashboard liefert auf einen Blick den Compliance-Status mit klaren Eskalationsausloesern. Bei Fehler: Koennen Datenquellen keine automatisierten Metriken unterstuetzen, manuelle Erfassung implementieren und den Plan zur Automatisierung dokumentieren.

Schritt 5: Untersuchungsausloeser und Eskalation etablieren

# Investigation and Escalation Matrix

### Investigation Triggers
| Trigger | Severity | Response Time | Investigator |
|---------|----------|---------------|-------------|
| Audit trail gap detected | Critical | Immediate (within 4 hours) | IT + QA |
| Confirmed data falsification | Critical | Immediate (within 4 hours) | QA Director |
| Anomaly pattern confirmed after review | Major | Within 5 business days | QA Analyst |
| Repeated anomalies from same user | Major | Within 5 business days | QA + HR |
| Overdue audit trail review | Minor | Within 10 business days | QA Manager |

### Escalation Path
| Level | Escalated To | When |
|-------|-------------|------|
| 1 | System Owner | Any confirmed anomaly |
| 2 | QA Director | Major or critical finding |
| 3 | Site Director | Critical finding or potential regulatory impact |
| 4 | Regulatory Affairs | Confirmed data integrity failure requiring regulatory notification |

Erwartet: Jede Untersuchung hat definierte Schwere, Zeitplan und Eskalationspfad. Bei Fehler: Werden Untersuchungen nicht innerhalb der definierten Zeitplaene abgeschlossen, auf die naechste Ebene eskalieren.

Schritt 6: Ueberwachungsplan zusammenstellen

Alle Komponenten in den Master-Datenintegritaets-Ueberwachungsplan zusammenfuehren:

# Data Integrity Monitoring Plan
## Document ID: DI-MONITORING-PLAN-[SITE]-[YYYY]-[NNN]

### 1. Purpose and Scope
[From assessment scope]

### 2. ALCOA+ Assessment Summary
[From Step 1]

### 3. Detective Controls
[From Step 2]

### 4. Anomaly Detection Rules
[From Step 3]

### 5. Metrics and Reporting
[From Step 4]

### 6. Investigation and Escalation
[From Step 5]

### 7. Periodic Review
- Monitoring plan review: Annual
- Anomaly thresholds: Adjust after each quarterly review
- ALCOA+ re-assessment: When systems change or new systems are added

### 8. Approval
| Role | Name | Signature | Date |
|------|------|-----------|------|
| QA Director | | | |
| IT Director | | | |
| Site Director | | | |

Erwartet: Ein einzelnes genehmigtes Dokument, das das vollstaendige Datenintegritaets-Ueberwachungsprogramm definiert. Bei Fehler: Ist der Plan fuer ein einzelnes Dokument zu gross, einen Masterplan mit Referenzen auf systemspezifische Ueberwachungsverfahren erstellen.

Validierung

  • ALCOA+-Bewertung fuer alle Systeme im Umfang abgeschlossen
  • Auditpfad-Pruefungsplan mit Haeufigkeit, Umfang und verantwortlichem Pruefer definiert
  • Mindestens 5 Anomalieerkennungsmuster mit spezifischen Schwellenwerten definiert
  • Metrik-Dashboard hat KPIs mit Gruen/Gelb/Rot-Schwellenwerten
  • Untersuchungsausloeser mit Schwere und Reaktionszeitplaenen definiert
  • Eskalationsmatrix erreicht Regulatory Affairs bei kritischen Befunden
  • Ueberwachungsplan von QA und IT-Leitung genehmigt
  • Regelmaessiger Pruefungsrhythmus etabliert

Haeufige Stolperfallen

  • Ueberwachung ohne Handlung: Metriken sammeln aber Anomalien nie untersuchen gibt ein falsches Sicherheitsgefuehl und ist schlimmer als keine Ueberwachung (es erzeugt Nachweise ignorierter Befunde).
  • Statische Schwellenwerte: Schwellenwerte basierend auf Vermutungen statt Basisdaten erzeugen uebermaeig viele Fehlalarme, die zu Alarmmudigkeit fuehren.
  • Auditpfad-Pruefung als Checkboxen-Uebung: Auditpfade ohne Verstaendnis, wonach man sucht, zu pruefen ist wirkungslos. Pruefer in Anomalieerkennungsmustern schulen.
  • Systemlimitierungen ignorieren: Manche Systeme haben schlechte Auditpfad-Faehigkeiten. Einschraenkungen dokumentieren und kompensierende Kontrollen implementieren, anstatt so zu tun als gaebe es die Einschraenkung nicht.
  • Kein Trending: Einzelne Anomalien koennen geringfuegig erscheinen, aber Muster ueber Zeit oder Nutzer hinweg offenbaren systemische Probleme. Datenintegritaetsmetriken immer im Trend verfolgen.

Verwandte Skills

  • design-compliance-architecture — identifiziert Systeme, die Datenintegritaets-Ueberwachung benoetigen
  • implement-audit-trail — die technische Grundlage, auf die Ueberwachung angewiesen ist
  • investigate-capa-root-cause — wenn Ueberwachung Probleme aufdeckt, die formale Untersuchung erfordern
  • conduct-gxp-audit — Audits bewerten die Wirksamkeit des Ueberwachungsprogramms
  • prepare-inspection-readiness — Datenintegritaet ist ein primaerer Schwerpunktbereich bei Behoerdeninspektionen

GitHub 저장소

pjt222/agent-almanac
경로: i18n/de/skills/monitor-data-integrity
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams
FAQ

Frequently asked questions

What is the monitor-data-integrity skill?

monitor-data-integrity is a Claude Skill by pjt222. Skills package instructions and resources that Claude loads on demand, so Claude can perform monitor-data-integrity-related tasks without extra prompting.

How do I install monitor-data-integrity?

Use the install commands on this page: add monitor-data-integrity to Claude Code as a plugin, or clone its repository into your skills directory, then restart Claude so it picks up the skill.

What category does monitor-data-integrity belong to?

monitor-data-integrity is in the Other category, tagged data.

Is monitor-data-integrity free to use?

Yes. monitor-data-integrity is listed on AIMCP and free to install. It runs inside Claude, so no separate service account is required to use the skill itself.

연관 스킬

llamaguard
기타

LlamaGuard는 폭력 및 혐오 발언 등 6가지 안전 범주에서 LLM 입력과 출력을 조정하기 위한 Meta의 70-80억 파라미터 모델입니다. 94-95% 정확도를 제공하며 vLLM, Hugging Face 또는 Amazon SageMaker를 사용해 배포할 수 있습니다. 이 기술을 사용하여 AI 애플리케이션에 콘텐츠 필터링 및 안전 가드레일을 손쉽게 통합하세요.

스킬 보기
cost-optimization
기타

이 Claude Skill은 리소스 적정화, 태깅 전략, 지출 분석을 통해 개발자들이 클라우드 비용을 최적화할 수 있도록 지원합니다. AWS, Azure, GCP에서 클라우드 비용을 절감하고 비용 거버넌스를 구현하기 위한 프레임워크를 제공합니다. 인프라 비용을 분석하거나, 리소스를 적정화하거나, 예산 제약을 충족해야 할 때 사용하세요.

스킬 보기
sports-betting-analyzer
기타

이 Claude Skill은 스프레드, 오버/언더, 프로프 베트를 포함한 스포츠 베팅 시장을 분석합니다. 역사적 추이와 상황별 통계를 검토하여 가치 베트를 발견하고, 교육적 목적으로 실행 가능한 권장 사항이 담긴 구조화된 마크다운 결과를 제공합니다. 개발자는 이 기능을 스포츠 베팅 분석 도구에 활용할 수 있으며, 단순히 엔터테인먼트/교육 목적으로만 설계되었음을 유의해야 합니다.

스킬 보기
quantizing-models-bitsandbytes
기타

이 스킬은 bitsandbytes를 사용하여 LLM을 8비트 또는 4비트 정밀도로 양자화하며, 최소한의 정확도 손실로 50-75%의 메모리 감소를 달성합니다. 제한된 GPU 메모리에서 더 큰 모델을 실행하거나 추론을 가속화하는 데 이상적이며, INT8, NF4, FP4와 같은 형식을 지원합니다. 이 스킬은 HuggingFace Transformers와 통합되어 QLoRA 학습 및 8비트 옵티마이저를 가능하게 합니다.

스킬 보기