Automated summary
Orchestrates the PDCA cycle (Plan-Do-Check-Act) for a software project using a network of specialized AI agents with bidirectional feedback loops.
Strengths
- Automates complex multi-agent workflows
- Structured phases with document generation
- Built-in validation loops (CTO, PO, Tech Lead, etc.)
- Clear separation of intent (what/why) from implementation (how) in documents
Limitations
- Requires detailed project configuration via bkit.config.json
- May be overkill for simple or quick changes
- Relies on specific agent mappings that must be maintained
When managing complex feature development or improvement cycles requiring multi-agent collaboration and structured feedback.
For minor or non-critical changes where a single agent approach suffices.
Security analysis
CautionThe skill is a PDCA workflow manager that legitimately requires Bash, Task, and file write operations. It does not instruct destructive or exfiltrating actions, but the presence of powerful execution tools warrants caution.
- •Uses Bash tool, which could execute arbitrary shell commands if the skill is invoked with untrusted input or if the instructions are manipulated.
- •Uses Task tool to invoke downstream agents, which may have their own capabilities; the chain of tool usage could escalate privileges.
- •Skill writes files and modifies `.pdca-status.json`, which could affect project tracking if misused.
Examples
pdca Implement user authenticationplan Add payment processingstatusname: pdca description: | Project-level PDCA skill override for gaegulzip. Integrates with project agents (CTO, PO, Tech Lead, etc.) and uses platform-specific document paths from bkit.config.json.
All agent interactions are BIDIRECTIONAL — downstream agents can push back to upstream agents when issues are found, with structured feedback loops.
Agent workflow per phase (↔ = bidirectional):
- Research: clarify ↔ research-director → CTO 통합 평가 (아키텍처+복잡도, BLOCK → clarify 재호출)
- Plan: PO ↔ CTO (타당성 스캔, BLOCK → PO 재호출) → 사용자 승인 → CTO 라우팅 ↔ PO (Scope Mismatch → PO 보완 + 재승인)
- Design: ui-ux-designer ↔ tech-lead (Design Pushback 루프) + API Contract 양방향 검증 (server/frontend tech-lead)
- Do: CTO (work-plan) → developers ↔ CTO (BLOCKED:QUESTIONS → 라우팅 → 답변 → 재개)
- Analyze: gap-detector ↔ CTO (integration review)
- Iterate/Report: bkit agents
Triggers: pdca, research, plan, design, analyze, report, status, next, iterate argument-hint: "[action] [feature]" user-invocable: true agents: research-brainstorm: clarify (Skill) research-director: bkit/research-director research-feasibility-cto: cto (Feasibility Council)
plan: product-owner plan-review: interactive-review:review (Skill) feasibility: cto routing: cto design-server: server/tech-lead design-mobile-ui: mobile/ui-ux-designer design-mobile-visual: frontend-design (Skill) design-mobile-tech: mobile/tech-lead design-web-ui: web/ui-ux-designer design-web-visual: frontend-design (Skill) design-web-tech: web/tech-lead do-distribute: cto do-server: server/node-developer do-mobile: mobile/flutter-developer do-web: web/react-developer test-scenario: test-scenario-generator design-review: plan-review (Skill, MANDATORY) analyze: bkit:gap-detector code-review: plan-review (Skill, MANDATORY) review: cto iterate: bkit:pdca-iterator report: bkit:report-generator verify: independent-reviewer allowed-tools:
- Read
- Write
- Edit
- Glob
- Grep
- Bash
- Task
- TaskCreate
- TaskUpdate
- TaskList
- AskUserQuestion imports:
- ${PLUGIN_ROOT}/templates/plan.template.md
- ${PLUGIN_ROOT}/templates/design.template.md
- ${PLUGIN_ROOT}/templates/do.template.md
- ${PLUGIN_ROOT}/templates/analysis.template.md
- ${PLUGIN_ROOT}/templates/report.template.md
- ${PLUGIN_ROOT}/templates/iteration-report.template.md next-skill: null pdca-phase: null task-template: "[PDCA] {feature}"
PDCA Skill (gaegulzip Project Override)
Unified Skill for managing PDCA cycle with agent-integrated workflow. Each PDCA phase automatically invokes the appropriate project agents from
bkit.config.json. Phase 상세 절차는phases/디렉토리의 개별 파일 참조.
Config Reference
Read bkit.config.json to determine:
context.defaultLanguage→ 모든 출력(문서, 상태 메시지, 에이전트 응답)에 이 언어 사용pdca.planDocPaths→ document output pathspdca.designDocPaths→ document output pathspdca.analyzeDocPaths→ analysis document output pathspdca.reportDocPaths→ report document output pathsplatforms.{platform}.agents→ agent mapping per phasepdca.statusFile→.pdca-status.jsonpathpdca.matchRateThreshold→ 90 (default)pdca.maxIterations→ 5 (default)
Language Rule
All output MUST be in the language specified by bkit.config.json → context.defaultLanguage.
When calling agents via Task tool, always append to the prompt:
\nIMPORTANT: Respond and write all documents in {defaultLanguage}. Code, paths, and technical terms stay in English.
⚠️ Intent-Only Document Rule (Plan & Design 문서 필수 원칙)
Plan/Design 문서는 "무엇을(WHAT)"과 "왜(WHY)"만 작성한다. "어떻게(HOW)" 즉 구현은 절대 작성하지 않는다.
이 원칙을 위반하는 예시:
- ❌ 파일명/경로 지정:
src/services/authService.ts에 구현 - ❌ 코드 구조 지시:
useEffect로 상태를 관리하고 Redux store에 저장 - ❌ 함수 시그니처:
async function validateToken(token: string): Promise<boolean> - ❌ DB 쿼리 작성:
SELECT * FROM users WHERE email = ? - ❌ 특정 라이브러리 사용법:
zod.object({ email: z.string().email() })로 검증
이 원칙을 지키는 예시:
- ✅ 사용자 행동 정의:
사용자는 이메일과 비밀번호로 로그인할 수 있다 - ✅ 비즈니스 규칙:
비밀번호는 최소 8자, 영문+숫자+특수문자 포함 - ✅ 데이터 요구사항:
사용자 프로필에는 닉네임, 프로필 이미지, 가입일이 포함된다 - ✅ API 계약 (Design):
POST /auth/login → 요청: { email, password }, 응답: { token, user } - ✅ 제약조건:
토큰 만료 시간은 24시간
이유: 구현은 Do 단계의 개발자 에이전트가 결정한다. 설계 문서에 구현을 쓰면 에이전트 자율성을 제한하고, 코드 변경 시 문서 부패를 유발한다.
모든 Plan/Design phase 에이전트 프롬프트에 이 원칙을 포함시킬 것.
Product Naming Convention
{product}는 앱 이름을 그대로 사용합니다:
wowa— 메인 앱admin— 어드민 대시보드talmosang— 탈모상 AI 두피 분석
경로 예: docs/wowa/auth/user-story.md, docs/admin/dashboard/web-brief.md
Fullstack frontendType 제약
frontendType은 단일 값만 허용합니다 ("mobile" 또는 "web").
Mobile + Web 동시 개발이 필요한 경우, 별도 PDCA 사이클로 분리합니다:
- 예:
wowafeature = fullstack (frontendType: mobile),adminfeature = fullstack (frontendType: web)
Status Tracking
.pdca-status.json stores platform and phase per feature:
{
"features": {
"{feature}": {
"phase": "research | plan | design | do | analyze | iterate | completed | archived",
"platform": "server | mobile | web | fullstack",
"frontendType": "mobile | web",
"startedAt": "ISO timestamp",
"documents": {}
}
}
}
IMPORTANT:
platformis determined during Plan phase by CTO and reused in all subsequent phases.frontendTypeis set when platform isfullstackto distinguish Mobile vs Web frontend agents.
Arguments
| Argument | Description | Example |
|----------|-------------|---------|
| research [feature] | clarify → research-director | /pdca research user-auth |
| plan [feature] | PO → interactive-review → CTO routing | /pdca plan user-auth |
| design [feature] | Design docs (per platform) | /pdca design user-auth |
| do [feature] | CTO distribution → dev agents | /pdca do user-auth |
| analyze [feature] | Gap analysis + CTO review | /pdca analyze user-auth |
| iterate [feature] | Auto improvement | /pdca iterate user-auth |
| report [feature] | Completion report | /pdca report user-auth |
| archive [feature] | Archive documents | /pdca archive user-auth |
| cleanup [feature] | Cleanup archived | /pdca cleanup |
| status | Current status | /pdca status |
| next | Next phase guide | /pdca next |
Phase Details
각 phase 실행 시 해당 파일을 Read하여 상세 절차를 따릅니다.
| Phase | File | Summary |
|-------|------|---------|
| research | phases/research.md | clarify → research-director (Decision Points) → CTO 통합 평가 (아키텍처+복잡도, BLOCK↔clarify 재호출) |
| plan | phases/plan.md | clarify → PO (Research Gate) ↔ CTO 타당성 스캔 (BLOCK→PO 재호출) → 사용자 승인 → CTO 라우팅 ↔ PO (Scope Mismatch→보완+재승인) |
| design | phases/design.md | 플랫폼별 디스패치 + tech-lead ↔ designer Pushback 루프 + API Contract 양방향 검증 (Fullstack) |
| do | phases/do.md | 선행조건 검증 → 설계 리뷰 → CTO work-plan → 사용자 승인 → 실행 모드 → 에이전트 구현 ↔ BLOCKED:QUESTIONS 프로토콜 |
| analyze | phases/analyze.md | gap-detector → FINDINGS 구조화 → 코드 리뷰 → CTO 통합 리뷰 |
| iterate | phases/iterate.md | FINDINGS severity 순 에이전트 디스패치 또는 Ralph Loop 자율 수정 |
| report | phases/report.md | report-generator → 배포/유지보수/v2 가이드 → CHANGELOG |
| archive | phases/archive.md | 문서 아카이브 → status 업데이트 |
status (Status Check)
- Read
.pdca-status.json - Display: feature, phase, platform, matchRate, documents
- Visualize progress
📊 PDCA Status
─────────────────────────────
Feature: {feature}
Platform: {platform}
Phase: {phase}
Match Rate: {matchRate}%
─────────────────────────────
[Research] ✅ → [Plan] ✅ → [Design] ✅ → [Do] 🔄 → [Check] ⏳ → [Act] ⏳
next (Next Phase)
| Current | Next | Action | |---------|------|--------| | None | research (권장) 또는 plan | research는 선택적. 복잡한 기능은 research 권장, 단순 기능은 plan 직행 가능 | | research | plan | PO → interactive-review → CTO routing | | plan | design | Platform-based design agents | | design | do | CTO distribution → dev agents | | do | check | Gap analysis + CTO review | | check (<90%) | act | Auto-iterate | | check (>=90%) | report | Completion report | | report | archive | Archive documents |
Agent Integration Summary (All Bidirectional)
↔= 양방향 (피드백 경로 있음),──→= 단방향 (최종 출력) 참조:.claude/guide/agent-communication-protocol.md— 전 phase 양방향 통신 프로토콜
Research: clarify (user) ──→ research-director (research.md + Decision Points)
│
↓ CTO 통합 평가 (아키텍처 + 복잡도)
│
BLOCK ↔ clarify 재호출 (max 2회)
CONCERN → AskUserQuestion
│
↓ Decision Points → PO에게 전달
Plan: PO (user-story.md) ↔ CTO (⓪-pre 타당성 스캔)
│ │ BLOCK → PO 재호출 (max 2회)
│ ↓ PASS/WARN
interactive-review (사용자 승인) ──→ CTO (⓪ 플랫폼 라우팅)
│
Scope Mismatch ↔ PO 보완 + 사용자 재승인
↓ platform & frontendType stored
Design: ┌── Server: tech-lead (server-brief)
├── Mobile: ui-ux-designer ↔ tech-lead (Design Pushback 루프, max 2회)
│ └── frontend-design (Skill) 포함
└── Web: ui-ux-designer ↔ tech-lead (Design Pushback 루프, max 2회)
└── frontend-design (Skill) 포함
│ Fullstack = Server + frontend (by frontendType)
│ + API Contract 양방향 검증 (server tech-lead ↔ CTO ↔ frontend tech-lead)
│
Do: CTO (work-plan) ──→ developers (node/flutter/react)
│ ↔ BLOCKED:QUESTIONS → CTO 라우팅 → 답변자 → 재개
│ OR: Ralph Loop (단일 에이전트 자율 반복)
│
Analyze: gap-detector ↔ CTO (review)
│
Iterate: pdca-iterator (if < 90%) OR: Ralph Loop (FINDINGS 자율 수정)
│
Report: report-generator ──→ CHANGELOG
│
Verify: independent-reviewer (optional)
Bidirectional Communication Reference
전 phase의 양방향 통신 프로토콜 상세는 아래 문서를 참조합니다:
- 통신 프로토콜:
.claude/guide/agent-communication-protocol.md - phase별 상세:
phases/{phase}.md
CTO Involvement (5 Points)
| Phase | CTO Role | Output | 양방향 |
|-------|---------|--------|--------|
| Research (after research-director) | 아키텍처+복잡도 통합 평가 | PASS/CONCERN/BLOCK + complexity 판정 | BLOCK → clarify 재호출 |
| Plan (after PO, before user review) | ⓪-pre 타당성 스캔 | PASS/WARN/BLOCK 판정 | BLOCK → PO 재호출 |
| Plan (after user review) | ⓪ 플랫폼 라우팅 + Scope Mismatch 감지 | platform in status | Mismatch → PO 보완 + 재승인 |
| Do (before devs) | ② 작업 분배 + BLOCKED:QUESTIONS 라우팅 | work-plan.md (플랫폼별) | BLOCKED → 적절한 에이전트 라우팅 |
| Analyze (after gap) | ③ 통합 리뷰 | cto-review.md (플랫폼별) | — |
From the file to a first result
The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.
View the source and supporting filesThis download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.
1. Download the file
Download SKILL.mdFrom the root of a test project, save the file as .skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.
Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.
mkdir -p '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/pdca-cycle-management-with-integrated-agents-mm5t8a1i/download' --output '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md.part' && mv '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md.part' '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md'2. Check prerequisites
Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.
No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.
The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.
3. Load the skill
After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.
Claude Code
test -f '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md' && test -s '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/pdca' && cp -R '.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i' '.claude/skills/pdca'OpenSkills
OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.
Official OpenSkills installationopenskills install './.skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i'openskills read 'pdca-cycle-management-with-integrated-agents-mm5t8a1i'Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.
For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.
openskills syncManual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.
4. Try a first task
Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:
Read .skill-downloads/pdca-cycle-management-with-integrated-agents-mm5t8a1i/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.
React Hook Form Best Practices
Development
Master React Hook Form: avoid anti-patterns, manage subscriptions, and handle reset, dirty state, and controlled inputs correctly.
Emotion to Mantine Migration
Development
Migrate Emotion styled-components to Mantine components with style props and CSS modules. Use when converting .styled.tsx files or removing @emotion imports.
Scope Creep Detector
Development
Analyzes git diffs against a stated intent to detect scope creep, unrelated files, broad PRs, and oversized changes, then suggests keep, split, or justify actions.