buildcrew 1.0.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 z1nun
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.ko.md

@@ -0,0 +1,236 @@
+# buildcrew
+
+Claude Code를 위한 11개 AI 에이전트 — 9개 운영 모드로 자동 오케스트레이션되는 개발 팀.
+
+```bash
+npx buildcrew
+```
+
+> [English](README.md) | **한국어**
+
+---
+
+## 왜 buildcrew인가?
+
+AI 코딩 에이전트는 강력하지만 구조 없이는 일관성 없는 결과를 냅니다. buildcrew는 Claude Code에 다음을 제공합니다:
+
+- **팀** — 명확한 역할을 가진 11개 전문 에이전트
+- **프로세스** — 품질 게이트와 반복을 갖춘 순차 파이프라인
+- **하네스** — 프로젝트 컨텍스트, 규칙, 도메인 지식이 모든 에이전트에 내장
+- **오케스트레이터** — `@constitution`에게 말하면 자동으로 적절한 에이전트 배치
+
+```
+나:     @constitution 유저 대시보드 추가해줘
+크루:   기획자 → 디자이너 → 개발자 → QA → 브라우저 QA → 리뷰어 → 배포
+```
+
+외부 의존성 없음. 런타임 없음. 바이너리 없음. 그냥 마크다운.
+
+---
+
+## 시작하기
+
+```bash
+# 1. 에이전트 설치
+npx buildcrew
+
+# 2. 프로젝트 하네스 설정 (인터랙티브)
+npx buildcrew init
+
+# 3. 추가 컨텍스트 (선택)
+npx buildcrew add erd
+npx buildcrew add architecture
+npx buildcrew add design-system
+
+# 4. 사용
+@constitution 유저 대시보드 추가해줘
+```
+
+---
+
+## 하네스 엔지니어링
+
+하네스는 "범용 AI 출력"과 "내 프로젝트에 맞는 코드"의 차이를 만듭니다.
+
+```bash
+npx buildcrew init
+```
+
+인터랙티브 설정으로 핵심 하네스 파일 생성:
+
+```
+[1/3] 프로젝트 컨텍스트
+  프로젝트명, 설명, 기술 스택 (자동 감지), 배포 타겟, 프로덕션 URL
+
+[2/3] 팀 규칙
+  코딩 컨벤션, 우선순위, 금지사항, 품질 기준, 리뷰 규칙
+
+[3/3] 도메인 지식
+  업종, 유저 타입, 도메인 용어, 비즈니스 룰
+```
+
+### 템플릿으로 추가 컨텍스트 제공
+
+```bash
+npx buildcrew add erd              # DB 스키마 & 관계
+npx buildcrew add architecture     # 시스템 아키텍처
+npx buildcrew add api-spec         # API 엔드포인트 & 계약
+npx buildcrew add design-system    # 색상, 타이포, 컴포넌트
+npx buildcrew add glossary         # 도메인 용어 & 유저 역할
+npx buildcrew add user-flow        # 유저 여정 & 페이지 맵
+npx buildcrew add env-vars         # 환경변수 가이드
+```
+
+### 열린 구조
+
+템플릿에 제한되지 않습니다. `.claude/harness/`에 아무 `.md` 파일이나 추가하세요:
+
+```
+.claude/harness/
+├── project.md          ← 핵심 (init으로 생성)
+├── rules.md            ← 핵심 (init으로 생성)
+├── erd.md              ← 템플릿 (add로 추가)
+├── architecture.md     ← 템플릿 (add로 추가)
+├── 내맘대로.md          ← 직접 작성 (에이전트가 읽음)
+└── 무엇이든.md          ← 아무 .md 파일 가능
+```
+
+### 하네스 파일 → 에이전트 라우팅
+
+| 파일 | 읽는 에이전트 |
+|------|-------------|
+| `project.md`, `rules.md` | 모든 에이전트 |
+| `erd.md`, `architecture.md`, `api-spec.md` | developer, reviewer, security-auditor, investigator |
+| `design-system.md` | designer |
+| `glossary.md`, `user-flow.md` | planner, designer, browser-qa |
+| `env-vars.md` | developer, security-auditor |
+| 커스텀 파일 | reviewer, security-auditor (전부 읽음) |
+
+### 하네스 관리
+
+```bash
+npx buildcrew harness          # 하네스 파일 상태 확인
+npx buildcrew add              # 사용 가능한 템플릿 목록
+npx buildcrew add erd          # 특정 템플릿 추가
+npx buildcrew init --force     # 핵심 파일 재생성
+```
+
+---
+
+## 에이전트
+
+### 빌드 팀
+
+| 에이전트 | 모델 | 역할 |
+|---------|------|------|
+| **planner** | opus | 6가지 강제 질문 + 4관점 자체 리뷰 (CEO, 엔지니어링, 디자인, QA). 관점별 1-10점 채점. |
+| **designer** | opus | 웹에서 UI/UX 레퍼런스 수집, Playwright로 실제 사이트 스크린샷, Figma MCP 통합, 프로덕션 React/Next.js 컴포넌트 작성. AI 슬롭 블랙리스트. |
+| **developer** | sonnet | 기획서 + 디자인 + 하네스 규칙에 따라 구현. |
+
+### 품질 팀
+
+| 에이전트 | 모델 | 역할 |
+|---------|------|------|
+| **qa-tester** | sonnet | 코드 레벨 검증 — 타입, 린트, 빌드, 수용 기준. |
+| **browser-qa** | sonnet | Playwright MCP 실제 브라우저 테스트 — 유저 플로우, 반응형, 콘솔, 건강 점수 (0-100). |
+| **reviewer** | opus | 4관점 병렬 리뷰 (보안, 성능, 테스트, 유지보수) + 적대적 리뷰 + 자동 수정. |
+| **health-checker** | sonnet | 코드 품질 대시보드 — 7개 카테고리 가중 점수 (0-10) + 트렌드. |
+
+### 보안 & 운영 팀
+
+| 에이전트 | 모델 | 역할 |
+|---------|------|------|
+| **security-auditor** | opus | OWASP Top 10 + STRIDE 위협 모델. 10단계 감사. |
+| **canary-monitor** | sonnet | 배포 후 프로덕션 헬스 — 페이지, API, 콘솔, 성능 비교. |
+| **shipper** | sonnet | 릴리즈 파이프라인 — 테스트 → 버전 → 체인지로그 → PR. |
+
+### 전문가
+
+| 에이전트 | 모델 | 역할 |
+|---------|------|------|
+| **investigator** | sonnet | 근본 원인 디버깅. 4단계 조사. 무관한 코드 수정 동결. |
+
+---
+
+## 9가지 운영 모드
+
+`@constitution`에게 자연스럽게 말하세요. 모드를 자동 감지합니다.
+
+| 모드 | 예시 | 파이프라인 |
+|------|------|----------|
+| **Feature** | "유저 대시보드 추가해줘" | 기획 → 디자인 → 개발 → QA → 브라우저 QA → 리뷰 |
+| **Project Audit** | "프로젝트 전체 점검해줘" | 스캔 → 우선순위 → 수정 → 검증 (반복) |
+| **Browser QA** | "브라우저 테스트해줘" | Playwright 테스트 + 건강 점수 |
+| **Security** | "보안 점검해줘" | OWASP + STRIDE + 시크릿 + 의존성 |
+| **Debug** | "왜 로그인이 안 돼?" | 4단계 근본 원인 조사 |
+| **Health** | "헬스체크 돌려줘" | 품질 대시보드 (타입, 린트, 의존성, i18n) |
+| **Canary** | "배포 확인해줘" | 프로덕션 모니터링 |
+| **Review** | "코드 리뷰해줘" | 멀티 전문가 + 적대적 + 자동 수정 |
+| **Ship** | "배포해줘" | 테스트 → 버전 → 체인지로그 → PR |
+
+### 모드 체이닝
+
+Constitution이 다음 모드를 자동 제안:
+- Feature 완료 → Ship 제안
+- Ship 완료 → Canary 제안
+- Canary CRITICAL → Debug 트리거
+
+---
+
+## 기능 파이프라인
+
+각 기능이 완전한 문서 체인을 생성:
+
+```
+.claude/pipeline/{기능명}/
+├── 01-plan.md           요구사항 + 4관점 리뷰 점수
+├── 02-references.md     실제 사이트에서 수집한 UI/UX 레퍼런스
+├── 02-design.md         디자인 결정 + 컴포넌트 스펙
+├── 03-dev-notes.md      구현 노트 + 변경 파일
+├── 04-qa-report.md      수용 기준 검증
+├── 05-browser-qa.md     건강 점수 + 스크린샷 + 플로우
+├── 06-review.md         리뷰 발견사항 + 자동 수정
+└── 07-ship.md           PR URL + 릴리즈 노트
+```
+
+---
+
+## CLI 레퍼런스
+
+| 명령어 | 설명 |
+|--------|------|
+| `npx buildcrew` | 11개 에이전트 설치 |
+| `npx buildcrew init` | 프로젝트 하네스 설정 (인터랙티브) |
+| `npx buildcrew init --force` | 하네스 재생성 |
+| `npx buildcrew add` | 사용 가능한 템플릿 목록 |
+| `npx buildcrew add <type>` | 하네스 템플릿 추가 |
+| `npx buildcrew harness` | 하네스 파일 상태 확인 |
+| `npx buildcrew --force` | 기존 에이전트 덮어쓰기 |
+| `npx buildcrew --list` | 에이전트 목록 (모델 포함) |
+| `npx buildcrew --uninstall` | 에이전트 제거 |
+| `npx buildcrew --version` | 버전 확인 |
+
+## 요구사항
+
+- **필수**: [Claude Code](https://claude.ai/code) CLI
+- **선택**: [Playwright MCP](https://github.com/anthropics/anthropic-quickstarts/tree/main/mcp-servers/playwright) — browser-qa, canary-monitor, designer용
+- **선택**: [Figma MCP](https://github.com/anthropics/anthropic-quickstarts/tree/main/mcp-servers/figma) — designer용
+
+```bash
+# 실제 브라우저 테스트 활성화
+claude mcp add playwright -- npx @anthropic-ai/mcp-server-playwright
+```
+
+## 커스터마이징
+
+모든 파일 수정 가능:
+
+```
+.claude/agents/      에이전트 정의 — 역할, 도구, 지시사항, 모델 수정
+.claude/harness/     프로젝트 컨텍스트 — 언제든 수정, 아무 .md 추가
+.claude/pipeline/    출력 문서 — 기능별 자동 생성
+```
+
+## 라이선스
+
+MIT

package/README.md

@@ -0,0 +1,276 @@
+# buildcrew
+
+> **English** | [한국어](README.ko.md)
+
+11 AI agents for Claude Code — auto-orchestrated dev team with 9 operating modes.
+
+```bash
+npx buildcrew
+```
+
+---
+
+## Why buildcrew?
+
+AI coding agents are powerful, but without structure they produce inconsistent results. buildcrew gives Claude Code:
+
+- **A team** — 11 specialized agents, each with a clear role
+- **A process** — sequential pipeline with quality gates and iteration
+- **A harness** — your project context, rules, and domain knowledge baked in
+- **An orchestrator** — just talk to `@constitution`, it routes automatically
+
+```
+You:   @constitution Add user authentication
+Crew:  Planner → Designer → Developer → QA → Browser QA → Reviewer → Ship
+```
+
+No external dependencies. No runtime. No binaries. Just Markdown.
+
+---
+
+## Getting Started
+
+```bash
+# 1. Install agents
+npx buildcrew
+
+# 2. Set up project harness (interactive)
+npx buildcrew init
+
+# 3. Add more context (optional)
+npx buildcrew add erd
+npx buildcrew add architecture
+npx buildcrew add design-system
+
+# 4. Start working
+@constitution Add user dashboard
+```
+
+---
+
+## Harness Engineering
+
+The harness is what separates "generic AI output" from "code that fits your project."
+
+```bash
+npx buildcrew init
+```
+
+Interactive setup creates core harness files:
+
+```
+[1/3] Project Context
+  Project name, description, tech stack (auto-detected), deploy target, production URL
+
+[2/3] Team Rules
+  Coding conventions, priorities, what to avoid, quality standards, review rules
+
+[3/3] Domain Knowledge
+  Industry, user types, domain terms, business rules
+```
+
+### Add more context with templates
+
+```bash
+npx buildcrew add erd              # Database schema & relationships
+npx buildcrew add architecture     # System architecture overview
+npx buildcrew add api-spec         # API endpoints & contracts
+npx buildcrew add design-system    # Colors, typography, components
+npx buildcrew add glossary         # Domain terms & user roles
+npx buildcrew add user-flow        # User journeys & page map
+npx buildcrew add env-vars         # Environment variables guide
+```
+
+### The harness is open
+
+Not limited to templates. Create any `.md` file in `.claude/harness/`:
+
+```
+.claude/harness/
+├── project.md          ← core (auto by init)
+├── rules.md            ← core (auto by init)
+├── erd.md              ← template
+├── architecture.md     ← template
+├── design-system.md    ← template
+├── glossary.md         ← template
+├── my-custom-notes.md  ← your own file (agents read it too)
+└── anything.md         ← any .md file works
+```
+
+All agents read harness files before every task. Constitution routes the right files to the right agents:
+
+| File | Routed to |
+|------|-----------|
+| `project.md`, `rules.md` | All agents |
+| `erd.md`, `architecture.md`, `api-spec.md` | developer, reviewer, security-auditor, investigator |
+| `design-system.md` | designer |
+| `glossary.md`, `user-flow.md` | planner, designer, browser-qa |
+| `env-vars.md` | developer, security-auditor |
+| Custom files | reviewer, security-auditor (read ALL) |
+
+### Manage the harness
+
+```bash
+npx buildcrew harness          # Show status of all harness files
+npx buildcrew add              # List available templates
+npx buildcrew add erd          # Add a specific template
+npx buildcrew init --force     # Regenerate core files from scratch
+```
+
+---
+
+## Agents
+
+### Build Team
+
+| Agent | Model | Role |
+|-------|-------|------|
+| **planner** | opus | 6 Forcing Questions + 4-Lens Self-Review (CEO, Engineering, Design, QA). Produces battle-tested plans scored 1-10 per lens. |
+| **designer** | opus | Searches web for UI/UX references, screenshots real sites with Playwright, integrates Figma MCP, writes production React/Next.js components. AI slop blacklist enforced. |
+| **developer** | sonnet | Implements features following plan + design + harness conventions. Self-reviews before handoff. |
+
+### Quality Team
+
+| Agent | Model | Role |
+|-------|-------|------|
+| **qa-tester** | sonnet | Code-level verification — types, lint, build, acceptance criteria check. |
+| **browser-qa** | sonnet | Real browser testing via Playwright MCP — user flows, responsive (375/768/1440px), console errors, network, health score (0-100). |
+| **reviewer** | opus | 4-specialist parallel review (security, performance, testing, maintainability) + adversarial pass + auto-fix. Fix-first approach. |
+| **health-checker** | sonnet | Code quality dashboard — 7-category weighted score (0-10), trend tracking, top 5 actionable items. |
+
+### Security & Ops
+
+| Agent | Model | Role |
+|-------|-------|------|
+| **security-auditor** | opus | OWASP Top 10 + STRIDE threat model. 10-phase audit. Secrets scan, dependency audit, AI/LLM security. Confidence gate (8/10 standard, 2/10 comprehensive). |
+| **canary-monitor** | sonnet | Post-deploy health — page availability, API status, console errors, performance vs baseline. |
+| **shipper** | sonnet | Release pipeline — pre-flight checks → version bump → changelog → commit → push → PR creation. |
+
+### Specialist
+
+| Agent | Model | Role |
+|-------|-------|------|
+| **investigator** | sonnet | Root cause debugging. 4 phases: investigate → hypothesize → test → fix. Edit freeze on unrelated code. "No fix without root cause." |
+
+---
+
+## 9 Operating Modes
+
+Talk to `@constitution` naturally. It auto-detects the mode.
+
+| Mode | Example | What happens |
+|------|---------|-------------|
+| **Feature** | "Add user dashboard" | Plan → Design → Dev → QA → Browser QA → Review |
+| **Project Audit** | "full project audit" | Scan → Prioritize → Fix → Verify (loop) |
+| **Browser QA** | "browser qa localhost:3000" | Playwright tests flows, responsive, console |
+| **Security** | "security audit" | OWASP + STRIDE + secrets + deps |
+| **Debug** | "debug: login broken" | 4-phase root cause investigation |
+| **Health** | "health check" | Quality dashboard (types, lint, deps, i18n) |
+| **Canary** | "canary https://myapp.com" | Post-deploy production monitoring |
+| **Review** | "code review" | Multi-specialist analysis + auto-fix |
+| **Ship** | "ship" | Test → version → changelog → PR |
+
+### Mode chaining
+
+Constitution auto-suggests the next mode:
+- Feature complete → Ship
+- Ship complete → Canary
+- Canary CRITICAL → Debug
+
+---
+
+## Feature Pipeline
+
+Each feature generates a full document chain:
+
+```
+.claude/pipeline/{feature}/
+├── 01-plan.md           Requirements + 4-lens review scores
+├── 02-references.md     Curated UI/UX references from real sites
+├── 02-design.md         Design decisions + component specs
+├── 03-dev-notes.md      Implementation notes + files changed
+├── 04-qa-report.md      Acceptance criteria verification
+├── 05-browser-qa.md     Health score + screenshots + flow results
+├── 06-review.md         Review findings + auto-fixes applied
+└── 07-ship.md           PR URL + release notes
+```
+
+---
+
+## CLI Reference
+
+| Command | Description |
+|---------|-------------|
+| `npx buildcrew` | Install 11 agents to `.claude/agents/` |
+| `npx buildcrew init` | Set up project harness (interactive) |
+| `npx buildcrew init --force` | Regenerate harness from scratch |
+| `npx buildcrew add` | List available harness templates |
+| `npx buildcrew add <type>` | Add a harness template (erd, architecture, etc.) |
+| `npx buildcrew harness` | Show status of all harness files |
+| `npx buildcrew --force` | Overwrite existing agent files |
+| `npx buildcrew --list` | List all agents with models |
+| `npx buildcrew --uninstall` | Remove installed agents |
+| `npx buildcrew --version` | Show version |
+
+## Requirements
+
+- **Required**: [Claude Code](https://claude.ai/code) CLI
+- **Optional**: [Playwright MCP](https://github.com/anthropics/anthropic-quickstarts/tree/main/mcp-servers/playwright) — for browser-qa, canary-monitor, designer
+- **Optional**: [Figma MCP](https://github.com/anthropics/anthropic-quickstarts/tree/main/mcp-servers/figma) — for designer
+
+```bash
+# Enable real browser testing
+claude mcp add playwright -- npx @anthropic-ai/mcp-server-playwright
+```
+
+## Customization
+
+Every file is editable:
+
+```
+.claude/agents/      Agent definitions — edit roles, tools, instructions, model
+.claude/harness/     Project context — edit anytime, add any .md file
+.claude/pipeline/    Output documents — auto-generated per feature
+```
+
+Change an agent's model:
+```yaml
+# In any agent .md file
+model: opus    # or sonnet, haiku
+```
+
+## Architecture
+
+```
+@constitution (orchestrator, opus)
+    │
+    ├─ reads .claude/harness/*.md (project context)
+    │
+    ├─ detects mode from user message
+    │
+    ├─ dispatches agents with relevant harness context
+    │
+    └─ enforces quality gates + iteration
+         │
+         ├── Build:    planner → designer → developer
+         ├── Quality:  qa-tester → browser-qa → reviewer
+         ├── Sec/Ops:  security-auditor, canary-monitor, shipper
+         └── Debug:    investigator
+```
+
+## Compared to gstack
+
+| | buildcrew | gstack |
+|---|-----------|--------|
+| Install | `npx buildcrew` | `git clone` + Bun + `./setup` |
+| Orchestration | Auto (constitution routes) | Manual (`/qa`, `/review`, ...) |
+| Harness | `init` + `add` + custom .md | Manual CLAUDE.md |
+| Dependencies | None | Bun, Playwright binary (~58MB) |
+| Browser testing | Playwright MCP | Custom Playwright daemon |
+| Pipeline docs | Auto-generated chain (01→07) | Per-skill output |
+| Agents | 11 (5 opus, 6 sonnet) | 34 |
+| Customization | Edit .md directly, never overwritten | May overwrite on update |
+
+## License
+
+MIT

package/agents/browser-qa.md

@@ -0,0 +1,130 @@
+---
+name: browser-qa
+description: Browser QA agent - performs real browser testing using Playwright MCP, captures screenshots, tests user flows, checks console errors, and verifies responsive design
+model: sonnet
+tools:
+  - Read
+  - Write
+  - Glob
+  - Grep
+  - Bash
+  - mcp__playwright__browser_navigate
+  - mcp__playwright__browser_click
+  - mcp__playwright__browser_fill_form
+  - mcp__playwright__browser_snapshot
+  - mcp__playwright__browser_take_screenshot
+  - mcp__playwright__browser_evaluate
+  - mcp__playwright__browser_console_messages
+  - mcp__playwright__browser_hover
+  - mcp__playwright__browser_press_key
+  - mcp__playwright__browser_wait_for
+  - mcp__playwright__browser_tabs
+  - mcp__playwright__browser_resize
+  - mcp__playwright__browser_network_requests
+  - mcp__playwright__browser_navigate_back
+  - mcp__playwright__browser_select_option
+  - mcp__playwright__browser_close
+  - mcp__playwright__browser_drag
+  - mcp__playwright__browser_file_upload
+  - mcp__playwright__browser_handle_dialog
+  - mcp__playwright__browser_run_code
+---
+
+# Browser QA Agent
+
+> **Harness**: Before starting, read `.claude/harness/project.md` and `.claude/harness/rules.md` if they exist. Follow all team rules defined there.
+
+
+You are a **Browser QA Tester** who performs real browser-based testing using Playwright. You actually navigate the application, click buttons, fill forms, and verify everything works from a real user's perspective.
+
+---
+
+## Test Tiers
+
+| Tier | Scope | When |
+|------|-------|------|
+| **Quick** | Affected pages only, happy paths | Small changes |
+| **Standard** | All major flows + edge cases | Feature completion (default) |
+| **Exhaustive** | Every page, every state, every breakpoint | Pre-release |
+
+---
+
+## Process
+
+### Phase 1: Setup & Orient
+1. Ensure dev server is running (check the provided URL or `http://localhost:3000`)
+2. If pipeline docs exist, read plan and dev notes to know what to verify
+3. Navigate to target URL, take initial snapshot
+4. Detect the application structure (routes, navigation, key pages)
+
+### Phase 2: Page Exploration
+For each page: navigate → snapshot → screenshot → check console → check network → identify interactive elements
+
+### Phase 3: User Flow Testing
+Test each flow end-to-end. After every interaction: check console for errors, verify expected outcome, screenshot key states.
+
+### Phase 4: State Testing
+For each interactive component verify: default, loading, error, empty, hover, active/focus, disabled states.
+
+### Phase 5: Responsive Testing
+Test at three breakpoints by resizing:
+- Mobile: 375 x 812
+- Tablet: 768 x 1024
+- Desktop: 1440 x 900
+
+### Phase 6: Accessibility Quick Check
+- Keyboard navigation: Tab through all interactive elements
+- Focus indicators visible?
+- ARIA labels present in accessibility tree?
+
+### Phase 7: Console & Network Audit
+Collect all console errors, check for 4xx/5xx API responses, CORS issues, failed resource loads.
+
+---
+
+## Health Score
+
+| Category | Weight |
+|----------|--------|
+| Console Errors | 15% |
+| Functional (flows) | 25% |
+| UX (states) | 20% |
+| Responsive | 15% |
+| Accessibility | 10% |
+| Performance | 10% |
+| Network Errors | 5% |
+
+Score: 90-100 Excellent, 70-89 Good, 50-69 Needs Work, <50 Critical.
+
+---
+
+## Output
+
+Write to `.claude/pipeline/{feature-name}/05-browser-qa.md`:
+
+```markdown
+# Browser QA Report: {Feature Name}
+## Test Configuration
+## Health Score: [NN]/100
+| Category | Score | Details |
+## Flows Tested
+| # | Flow | Status | Notes |
+## Issues Found
+### ISSUE-NNN: [Title]
+- Severity, Category, Page, Steps to Reproduce, Expected, Actual, Suggested Fix
+## Console Errors
+## Responsive Results
+## Overall Status: [PASS | FAIL | PARTIAL]
+## Verdict: [SHIP / FIX REQUIRED]
+```
+
+---
+
+## Rules
+1. Always screenshot before and after key interactions
+2. Always check console after every navigation and major interaction
+3. Test like a user, not a developer
+4. Don't guess — actually click it, actually resize
+5. Be specific in bug reports
+6. Test the unhappy path — what happens when things go wrong?
+7. Mobile first — test smallest screen first