Skip to content

pdca-methodology.md

Latest commit

 

History

History
223 lines (164 loc) · 7.52 KB

pdca-methodology.md

File metadata and controls

223 lines (164 loc) · 7.52 KB

PDCA Methodology in bkit

PDCA 사이클과 9단계 파이프라인의 관계

PDCA Cycle

┌─────────────────────────────────────────────────────────────────┐
│                      PDCA Cycle                                  │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│   Plan ──────► Design ──────► Do ──────► Check ──────► Act     │
│     │                                                    │      │
│     └────────────────── Improvement Cycle ◄──────────────┘      │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Phases

Phase Document Location Command Purpose
Plan docs/01-plan/ /pdca-plan 목표, 범위, 성공 기준 정의
Design docs/02-design/ /pdca-design 아키텍처, 데이터 모델, API 스펙
Do Code - 설계 기반 구현
Check docs/03-analysis/ /pdca-analyze 설계-구현 Gap 분석
Act docs/04-report/ /pdca-report 완료 보고서, 학습 사항 정리

9-Stage Development Pipeline

전체 프로젝트 개발을 위한 9단계 파이프라인:

Phase 1: Schema       → Data modeling, terminology
Phase 2: Convention   → Coding conventions
Phase 3: Mockup       → UI/UX mockups
Phase 4: API          → API design and implementation
Phase 5: Design System → Build design system
Phase 6: UI           → UI component integration
Phase 7: SEO/Security → SEO and security checks
Phase 8: Review       → Code review
Phase 9: Deployment   → Production deployment

Level-specific Flow

Level Phases Notes
Starter 1 → 2 → 3 → 6 → 9 Phase 4, 5, 7, 8 skip
Dynamic 1 → 2 → 3 → 4 → 5 → 6 → 7 → 9 Phase 8 optional
Enterprise All phases All phases required

PDCA vs Pipeline: The Key Relationship

Core Concept

┌─────────────────────────────────────────────────────────────────┐
│                                                                 │
│   "Run PDCA cycle WITHIN each Pipeline Phase"                   │
│                                                                 │
│   NOT: Pipeline as a whole = PDCA                               │
│   YES: Each Phase = One PDCA cycle                              │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Example: Phase 4 (API)

Phase 4: API Implementation

├── Plan: Define API endpoints and requirements
├── Design: Write API spec (OpenAPI/REST)
├── Do: Implement endpoints
├── Check: Zero Script QA + Gap analysis
└── Act: Document learnings, fix issues

Auto-Apply Rules

bkit-rules skill이 자동으로 PDCA 적용:

1. Task Classification

코드 변경 시 자동 분류:

Quick Fix      (< 10 lines)   → PDCA 선택적
Minor Change   (< 50 lines)   → PDCA 권장
Feature        (< 200 lines)  → PDCA 필수
Major Feature  (>= 200 lines) → PDCA + 분할 권장

2. Design Document Check

Write/Edit 시 (PreToolUse hook):

1. 해당 기능의 design doc 존재 확인
2. 없으면 → "설계 먼저 작성할까요?" 제안
3. 있으면 → 설계 참조하여 구현

3. Gap Analysis Suggestion

구현 완료 후 (PostToolUse hook):

1. 변경 파일 감지
2. Feature 크기 분석
3. "Gap Analysis 실행할까요?" 제안

Document Templates

Template Purpose Key Sections
plan.template.md 계획 문서 Goals, Scope, Success Criteria, Schedule
design.template.md 설계 문서 Architecture, Data Model, API Spec, Test Plan
design-starter.template.md Starter용 간소화 설계 Goals, User Flow, Components
design-enterprise.template.md Enterprise용 상세 설계 Service Architecture, Infrastructure
analysis.template.md Gap 분석 문서 Match Rate, Missing/Added/Changed, Recommendations
report.template.md 완료 보고서 Completed Items, Learnings, Improvements

Archive Rules

PDCA 사이클 완료 후 archive로 이동:

Archive Trigger

Condition Action
Gap Analysis >= 90% match Archive 가능
Report 생성 완료 Archive 권장
사용자 명시적 완료 선언 Archive 실행

Archive Process

1. 완료 조건 확인
   └── docs/03-analysis/{feature}.analysis.md 존재
   └── match rate >= 90% 또는 사용자 승인

2. Archive 폴더 생성
   └── docs/archive/YYYY-MM/{feature}/

3. 관련 문서 이동
   ├── docs/pdca/01-plan/features/{feature}.plan.md → archive
   ├── docs/pdca/02-design/features/{feature}.design.md → archive
   └── docs/pdca/03-analysis/{feature}.analysis.md → archive

4. 인덱스 업데이트
   └── docs/archive/YYYY-MM/_INDEX.md에 추가

Zero Script QA

전통적인 테스트 스크립트 없이 QA를 수행하는 방법론:

Traditional vs Zero Script

Aspect Traditional QA Zero Script QA
Setup Write test scripts Build log infrastructure (one-time)
Execution Run scripts Manual UX testing
Analysis Check results AI real-time log analysis
Maintenance Update scripts per change No maintenance needed

Core Principles

1. Log everything (including 200 OK)
2. Structured JSON logs
3. Track entire flow with Request ID
4. AI monitors real-time and documents issues

Commands Reference

PDCA Workflow

Command Function
/pdca-status Current PDCA progress dashboard
/pdca-plan [feature] Write plan document
/pdca-design [feature] Write design document
/pdca-analyze [feature] Gap analysis (design vs implementation)
/pdca-iterate [feature] Auto-fix with Evaluator-Optimizer pattern
/pdca-report Generate completion report
/pdca-next Suggest next action

Pipeline Management

Command Function
/pipeline-start Start full development pipeline
/pipeline-status Check pipeline status
/pipeline-next Next pipeline phase

Related Documents

  • [[core-mission]] - 핵심 사명과 철학
  • [[ai-native-principles]] - AI-Native 원칙
  • [[../components/skills/_skills-overview]] - Skill 상세
  • [[../scenarios/scenario-new-feature]] - 새 기능 시나리오
  • [[../../skills/development-pipeline/SKILL]] - 파이프라인 skill