关于
This skill performs Computer Systems Validation (CSV) assessments using the GAMP 5 methodology for GxP-regulated environments. It handles user requirements, risk analysis, IQ/OQ/PQ planning, traceability matrices, and validation reporting. Use it when introducing a new system, making major changes to a validated system, during periodic revalidation, or for regulatory inspection gap analyses.
快速安装
Claude Code
推荐npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/perform-csv-assessment在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
行 CSV 評
依 GAMP 5 險基法為受監境之計算系驗。
用
- GxP 境引新系
- 舊驗系大變
- 期重驗
- 巡備求驗隙析
入
- 必:系述(名、用、商、版)
- 必:用旨+監境(GxP 範)
- 必:GAMP 5 軟類(1-5)
- 可:現 URS
- 可:商文(設、釋、SOP)
- 可:舊驗文
行
一:定 GAMP 5 軟類
分類:
| Category | Type | Example | Validation Effort |
|---|---|---|---|
| 1 | Infrastructure software | OS, firmware | Low — verify installation |
| 3 | Non-configured product | COTS as-is | Low-Medium — verify functionality |
| 4 | Configured product | LIMS with config | Medium-High — verify configuration |
| 5 | Custom application | Bespoke R/Shiny app | High — full lifecycle validation |
得: 類明賦+因錄。 敗: 類歧→默高類+錄因。
二:書 URS
立 URS 文+號求:
# User Requirements Specification
## System: [System Name] v[Version]
## Document ID: URS-[SYS]-[NNN]
### 1. Purpose
[Intended use statement]
### 2. Functional Requirements
| ID | Requirement | Priority | Source |
|----|-------------|----------|--------|
| URS-001 | System shall calculate BMI from height and weight inputs | Must | Regulatory SOP-xxx |
| URS-002 | System shall generate audit trail entries for all data changes | Must | 21 CFR 11.10(e) |
| URS-003 | System shall export results in PDF format | Should | User request |
### 3. Non-Functional Requirements
| ID | Requirement | Priority | Source |
|----|-------------|----------|--------|
| URS-010 | System shall respond within 3 seconds for standard queries | Should | Usability |
| URS-011 | System shall restrict access via role-based authentication | Must | 21 CFR 11.10(d) |
### 4. Data Integrity Requirements
[ALCOA+ requirements: Attributable, Legible, Contemporaneous, Original, Accurate]
### 5. Regulatory Requirements
[Specific 21 CFR Part 11, EU Annex 11, or other applicable requirements]
得: 諸求皆有獨號、優、源。 敗: 無清源/優之求→旗以待干係人審。
三:行險評
施 GAMP 5 險基(FMEA):
# Risk Assessment
## Document ID: RA-[SYS]-[NNN]
| Req ID | Failure Mode | Severity (1-5) | Probability (1-5) | Detectability (1-5) | RPN | Risk Level | Mitigation |
|--------|-------------|----------------|-------------------|---------------------|-----|------------|------------|
| URS-001 | Incorrect BMI calculation | 4 | 2 | 1 | 8 | Low | OQ test case |
| URS-002 | Audit trail entries missing | 5 | 3 | 3 | 45 | High | IQ + OQ + monitoring |
| URS-011 | Unauthorized access | 5 | 2 | 2 | 20 | Medium | OQ test + periodic review |
RPN = 重 × 概 × 察。
| RPN Range | Risk Level | Testing Requirement |
|---|---|---|
| 1–12 | Low | Basic verification |
| 13–36 | Medium | Documented test case |
| 37+ | High | Full IQ/OQ/PQ with retest |
得: 各 URS 求皆有對應險評行。 敗: 未評之求→升驗主導前處。
四:定驗策(驗計)
# Validation Plan
## Document ID: VP-[SYS]-[NNN]
### Scope
- System: [Name] v[Version]
- GAMP Category: [N]
- Validation approach: [Prospective / Retrospective / Concurrent]
### Qualification Stages
| Stage | Scope | Applies? | Rationale |
|-------|-------|----------|-----------|
| IQ | Installation correctness | Yes | Verify installation, dependencies, configuration |
| OQ | Operational requirements | Yes | Verify functional requirements from URS |
| PQ | Performance under real conditions | [Yes/No] | [Rationale] |
### Roles and Responsibilities
| Role | Name | Responsibility |
|------|------|---------------|
| Validation Lead | [Name] | Plan, coordinate, approve |
| Tester | [Name] | Execute test scripts |
| System Owner | [Name] | Approve for production use |
| QA | [Name] | Review and sign-off |
### Acceptance Criteria
- All critical test cases pass
- No unresolved critical or major deviations
- Traceability matrix complete
得: 試行前諸干係人批驗計。 敗: 計未批勿行試。
五:立試程(IQ/OQ/PQ)
各驗階寫試本:
# Operational Qualification Protocol
## Test Case: TC-OQ-001
## Traces to: URS-001
**Objective:** Verify BMI calculation accuracy
**Prerequisites:**
- System installed per IQ protocol
- Test data set prepared
**Test Steps:**
| Step | Action | Expected Result | Actual Result | Pass/Fail |
|------|--------|-----------------|---------------|-----------|
| 1 | Enter height=180cm, weight=75kg | BMI displayed as 23.15 | | |
| 2 | Enter height=160cm, weight=90kg | BMI displayed as 35.16 | | |
| 3 | Enter height=0, weight=75kg | Error message displayed | | |
**Tester:** _________ Date: _________
**Reviewer:** _________ Date: _________
得: 諸中+高險求皆至少一試例。 敗: 行前補缺之試例。
六:立追溯陣
立 RTM、由各求經險評至試例:
# Traceability Matrix
## Document ID: TM-[SYS]-[NNN]
| URS ID | Requirement | Risk Level | Test Case(s) | Test Result | Status |
|--------|-------------|------------|--------------|-------------|--------|
| URS-001 | BMI calculation | Low | TC-OQ-001 | Pass | Verified |
| URS-002 | Audit trail | High | TC-IQ-003, TC-OQ-005 | Pass | Verified |
| URS-003 | PDF export | Low | TC-OQ-008 | Pass | Verified |
| URS-011 | Role-based access | Medium | TC-OQ-010, TC-OQ-011 | Pass | Verified |
得: 100% URS 求現於 RTM+鏈接試果。 敗: 任求無鏈試果→旗為驗隙。
七:書驗要報
# Validation Summary Report
## Document ID: VSR-[SYS]-[NNN]
### 1. Executive Summary
[System name] v[version] has been validated in accordance with [VP document ID].
### 2. Validation Activities Performed
| Activity | Document ID | Status |
|----------|-------------|--------|
| User Requirements | URS-SYS-001 | Approved |
| Risk Assessment | RA-SYS-001 | Approved |
| Validation Plan | VP-SYS-001 | Approved |
| IQ Protocol/Report | IQ-SYS-001 | Executed — Pass |
| OQ Protocol/Report | OQ-SYS-001 | Executed — Pass |
| Traceability Matrix | TM-SYS-001 | Complete |
### 3. Deviations
| Dev ID | Description | Impact | Resolution |
|--------|-------------|--------|------------|
| DEV-001 | [Description] | [Impact assessment] | [Resolution and rationale] |
### 4. Conclusion
The system meets all user requirements as documented in [URS ID]. The validation is considered [Successful / Successful with conditions].
### 5. Approval
| Role | Name | Signature | Date |
|------|------|-----------|------|
| Validation Lead | | | |
| System Owner | | | |
| Quality Assurance | | | |
得: 報引諸驗交付+清成/敗結。 敗: 偏差未決→報須言「條件」狀+CAPA 引。
驗
- GAMP 5 類賦+錄因
- URS 有號求+優+源追溯
- 險評涵諸 URS 求
- 試行前驗計批
- 試程有先決、步、預果、簽位
- RTM 鏈各求至險+試果
- 驗要報錄諸動、偏差、結
- 諸文皆有獨號+版控
忌
- 過驗:類 5 力施於類 3 軟費資—力依險合
- 缺追溯:求不貫至試例→隱隙
- 無計而試:批驗計前行試→果無效
- 忽非功能:安、性能、料整之求常忽
- 靜驗:視驗為一次—變則重評
參
setup-gxp-r-projectwrite-validation-documentationimplement-audit-trailvalidate-statistical-outputconduct-gxp-audit
GitHub 仓库
Frequently asked questions
What is the perform-csv-assessment skill?
perform-csv-assessment is a Claude Skill by pjt222. Skills package instructions and resources that Claude loads on demand, so Claude can perform perform-csv-assessment-related tasks without extra prompting.
How do I install perform-csv-assessment?
Use the install commands on this page: add perform-csv-assessment 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 perform-csv-assessment belong to?
perform-csv-assessment is in the Meta category, tagged design.
Is perform-csv-assessment free to use?
Yes. perform-csv-assessment is listed on AIMCP and free to install. It runs inside Claude, so no separate service account is required to use the skill itself.
相关推荐技能
Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。
这个Claude Skill为开发者提供完整的Polymarket预测市场开发支持,涵盖API调用、交易执行和市场数据分析。关键特性包括实时WebSocket数据流,可监控实时交易、订单和市场动态。开发者可用它构建预测市场应用、实施交易策略并集成实时市场预测功能。
该Skill帮助开发者创建OpenCode插件,用于接入命令、文件、LSP等25+种事件。它提供了插件结构、事件API规范和JavaScript/TypeScript实现模式,适合需要拦截操作、扩展功能或自定义事件处理的场景。开发者可通过它快速构建响应式模块来增强OpenCode AI助手的能力。
SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。
