Back to Skills

monitor-data-integrity

pjt222
Updated 6 days ago
17 views
17
2
17
View on GitHub
Designaidesigndata

About

This skill helps developers design and operate a data integrity monitoring program for GxP systems based on ALCOA+ principles. It provides detective controls, audit trail reviews, and anomaly detection for patterns like off-hours activity or bulk changes. Use it when establishing a compliance program, preparing for inspections, or implementing regulatory guidance from MHRA, WHO, or PIC/S.

Quick Install

Claude Code

Recommended
Primary
npx skills add pjt222/agent-almanac -a claude-code
Plugin CommandAlternative
/plugin add https://github.com/pjt222/agent-almanac
Git CloneAlternative
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/monitor-data-integrity

Copy and paste this command in Claude Code to install this skill

Documentation

監數整

立 ALCOA+ 之程:察、警、究。

  • GxP 系→立程
  • 巡視前→備
  • 事故後→加察
  • 控之週校
  • 行 MHRA/WHO/PIC/S 之導

  • :在範系+ALCOA+ 險像
  • :適導(MHRA、WHO TRS 996、PIC/S PI 041)
  • :各系現審跡能
  • :舊發現/巡觀
  • :現程/指
  • :權陣+角定

一:察 ALCOA+ 姿

各系對 ALCOA+ 諸原則:

# 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

得: 各系皆有評+具發現。 敗: 不能評→旗為要隙→即補。

二:設察控

定察違整之動:

# 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

得: 控有期、有人、有則。 敗: 過期未審→記隙→升 QA。

三:定異模

具觸究之模:

# 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

得: 模具、可量、可動。 敗: 太低(偽警繁)→依基線調;太高(漏實)→首巡後緊。

四:建指板

# 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 |

得: 板一覽合規+升觸。 敗: 源不能自動→人工+計劃。

五:立究與升

# 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 |

得: 各究有重、限、升。 敗: 逾限未畢→升一級。

六:成監計

合諸件為主計:

# 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 | | | |

得: 一已批之文定全程。 敗: 過大→主計+各系附程。

  • 諸範系皆評 ALCOA+
  • 審跡日表定(期、範、人)
  • 至少五異模+具閾
  • 板有 KPI+綠黃紅閾
  • 究觸有重+應期
  • 升路達監管事務(要事)
  • 計經 QA+IT 批
  • 週校期定

  • 察而不動:採指不究→偽安、生未理證據
  • 靜閾:猜閾→偽警繁→警疲
  • 審跡作勾:不識何察→無效→訓人辨模
  • 忽系限:跡能弱→錄限+補控、勿假裝
  • 無趨:個異微→跨時跨人成系疾→必趨

  • design-compliance-architecture
  • implement-audit-trail
  • investigate-capa-root-cause
  • conduct-gxp-audit
  • prepare-inspection-readiness

GitHub Repository

pjt222/agent-almanac
Path: i18n/wenyan-ultra/skills/monitor-data-integrity
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

Related Skills

executing-plans

Design

Use the executing-plans skill when you have a complete implementation plan to execute in controlled batches with review checkpoints. It loads and critically reviews the plan, then executes tasks in small batches (default 3 tasks) while reporting progress between each batch for architect review. This ensures systematic implementation with built-in quality control checkpoints.

View skill

requesting-code-review

Design

This skill dispatches a code-reviewer subagent to analyze code changes against requirements before proceeding. It should be used after completing tasks, implementing major features, or before merging to main. The review helps catch issues early by comparing the current implementation with the original plan.

View skill

connect-mcp-server

Design

This skill provides a comprehensive guide for developers to connect MCP servers to Claude Code using HTTP, stdio, or SSE transports. It covers installation, configuration, authentication, and security for integrating external services like GitHub, Notion, and custom APIs. Use it when setting up MCP integrations, configuring external tools, or working with Claude's Model Context Protocol.

View skill

web-cli-teleport

Design

This skill helps developers choose between Claude Code Web and CLI interfaces based on task analysis, then enables seamless session teleportation between these environments. It optimizes workflow by managing session state and context when switching between web, CLI, or mobile. Use it for complex projects requiring different tools at various stages.

View skill