Auto Tdd Executor

Auto TDD Executor v2.2

Automated TDD development with Ralph Wiggum loops, Beads persistence, Parallel Playwright E2E testing, Claude Code Hooks, and Code Simplifier integration.

What's New in v2.2

• Code Simplifier Agent: TDD REFACTOR 단계에서 자동 코드 정리
• Claude Code Hooks: Automatic quality gates, TDD cycle verification, subagent tracking
• Parallel E2E Testing: Group-based parallel test generation and healing
• Dependency Grouping: Analyze test scenarios and group by independence
• 10x Faster: Multiple Generator/Healer agents run concurrently
• Quality Gate Script: Automated validation with ./scripts/quality-gate.sh
• Simplified Prompts: 90+ lines → 30 lines for better reliability

🎯 What This Solves

Problem	Solution
Feature Planner requires manual execution	Ralph loops automate each phase
Ralph lacks structure	Feature plans provide quality gates
Long sessions lose context	Beads survives compaction
TDD is tedious manually	Automated Red-Green-Refactor cycles
E2E tests break often	Parallel Playwright agents auto-heal tests
Sequential test generation is slow	Parallel agents = 10x faster

🚀 Quick Start

# 1. Install prerequisites
/plugin install ralph-wiggum@anthropics
npm install -g @beads/bd
npm install -D @playwright/test
npx playwright install

# 2. Initialize project
bd init
/auto-tdd-start

# 3. Execute phases
/auto-tdd-run   # TDD with unit tests + E2E

# 4. Check status
/auto-tdd-status

🔄 Workflow

PLAN → TDD → E2E (Parallel) → QUALITY GATE → COMMIT → NEXT PHASE
        │           │
        │           ├─── PLANNER: Create test scenarios
        │           ├─── GROUPING: Analyze dependencies
        │           ├─── GENERATORS (Parallel): Create tests by group
        │           └─── HEALERS (Parallel): Fix failed tests
        │
        └─── 🔴 RED → 🟢 GREEN → 🔵 REFACTOR (tdd-refactor)

📋 Commands

Command	Description
/auto-tdd-start	Initialize project with Beads issues
/auto-tdd-run	Execute TDD phase (Red-Green-Refactor + E2E)
/auto-tdd-e2e	Run E2E tests with parallel Playwright agents
/auto-tdd-status	Show progress

🎭 Playwright Agents (Parallel Execution)

Agent	Purpose	Parallel?
playwright-test-planner	Explore app, create test scenarios	Single
playwright-test-generator	Generate test files by group	Parallel
playwright-test-healer	Fix failing tests	Parallel

Parallel E2E Workflow

1. PLANNER: Create test scenarios
      │
      ▼
2. GROUPING: Analyze dependencies, create independent groups
      │
      ├─────────────┬─────────────┬─────────────┐
      ▼             ▼             ▼             ▼
3. GENERATOR    GENERATOR     GENERATOR     GENERATOR
   (Group A)    (Group B)     (Group C)     (Group D)
   [Parallel]   [Parallel]    [Parallel]    [Parallel]
      │             │             │             │
      └─────────────┴─────────────┴─────────────┘
                    │
                    ▼
4. RUN ALL E2E TESTS
      │
      ├─── ALL PASS → Complete ✅
      │
      └─── SOME FAIL ──▶ 5. PARALLEL HEALERS
                              │
                   ┌──────────┼──────────┐
                   ▼          ▼          ▼
                HEALER     HEALER     HEALER
                (File1)    (File2)    (File3)
                [Parallel] [Parallel] [Parallel]

⚡ Performance Comparison

Approach	50 Tests Generation	10 Failed Tests Healing
Sequential	~50 min	~20 min
Parallel (5 agents)	~10 min	~4 min
Parallel (10 agents)	~5 min	~2 min

✅ Quality Gate

./scripts/quality-gate.sh --verbose

# Checks: Unit Tests, Lint, Type Check, Build, E2E
# Returns: QUALITY_GATE_PASSED or QUALITY_GATE_FAILED

🪝 Claude Code Hooks (v2.1)

Auto TDD Executor integrates with Claude Code v2.1.0 hooks for automated quality control:

Configured Hooks

Hook Event	Trigger	Action
PostToolUse	Edit/Write/MultiEdit	Quality Gate 자동 검증
PostToolUse	Bash (git commit)	Beads 상태 자동 업데이트
Stop	Agent 종료 전	TDD 사이클 완료 검증
SubagentStop	Generator/Healer 완료	병렬 에이전트 추적 및 로깅

Hook Scripts

hooks/
├── quality-gate-hook.sh      # 코드 수정 후 Quality Gate 실행
├── tdd-cycle-stop-hook.sh    # 테스트 통과 전 종료 차단
├── subagent-tracker-hook.sh  # 병렬 에이전트 완료 추적
├── beads-auto-update-hook.sh # git commit 후 beads 자동 업데이트
└── hooks.json                # Hooks 설정 파일

Hook Flow

┌─────────────────────────────────────────────────────────────────┐
│                     Hooks Integration                           │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Code Edit/Write                                                │
│       │                                                         │
│       ▼                                                         │
│  [PostToolUse Hook] ──▶ quality-gate-hook.sh                   │
│       │                    ├── Lint check                       │
│       │                    ├── Type check                       │
│       │                    └── Quick validation                 │
│       ▼                                                         │
│  git commit                                                     │
│       │                                                         │
│       ▼                                                         │
│  [PostToolUse Hook] ──▶ beads-auto-update-hook.sh              │
│       │                    └── Auto-close beads issue           │
│       ▼                                                         │
│  Agent tries to Stop                                            │
│       │                                                         │
│       ▼                                                         │
│  [Stop Hook] ──▶ tdd-cycle-stop-hook.sh                        │
│       │              ├── Run npm test                           │
│       │              ├── PASS → Allow stop                      │
│       │              └── FAIL → Block stop, continue TDD        │
│       ▼                                                         │
│  Subagent (Generator/Healer) completes                          │
│       │                                                         │
│       ▼                                                         │
│  [SubagentStop Hook] ──▶ subagent-tracker-hook.sh              │
│                              └── Log completion status          │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Manual Hook Testing

# Check registered hooks
/hooks

# Test hook scripts manually
echo '{"tool_name":"Write","tool_input":{"file_path":"test.ts"}}' | ./hooks/quality-gate-hook.sh

# Debug mode
claude --debug

📁 Files

auto-tdd-executor/
├── agents/                   # TDD Agents (v2.2)
│   └── tdd-refactor.md    # REFACTOR 단계 코드 정리
├── commands/
│   ├── start.md          # Initialize project
│   ├── run.md            # TDD execution (with tdd-refactor)
│   ├── e2e.md            # Parallel E2E testing
│   └── status.md         # Show progress
├── hooks/                    # Claude Code Hooks
│   ├── quality-gate-hook.sh
│   ├── tdd-cycle-stop-hook.sh
│   ├── subagent-tracker-hook.sh
│   ├── beads-auto-update-hook.sh
│   └── hooks.json
├── scripts/
│   ├── auto-tdd-runner.sh    # Automation script
│   └── quality-gate.sh       # Quality checks
├── skills/
│   └── auto-tdd-executor/SKILL.md
├── prompt-templates.md       # Simplified templates
└── README.md

📋 E2E Parallel Prompts

Generator Parallel Execution

"Playwright 테스트 제너레이터 에이전트를 활용해서
그룹별로 병렬로 에이전트를 띄워서 테스트 코드를 작성해줘.

Groups:
- Group A: [login, logout, password-reset]
- Group B: [product-list, product-search]
- Group C: [cart-add, cart-remove, checkout]

Use SINGLE message with MULTIPLE Task tool calls."

Healer Parallel Execution

"실패한 테스트 케이스에 대해서는 Playwright 테스트 힐러 에이전트 사용해서 수정해줘.
힐러 에이전트는 병렬로 띄워서 처리해줘.

Failed tests:
- tests/auth/login.spec.ts
- tests/products/search.spec.ts
- tests/cart/checkout.spec.ts

Each Healer runs independently in parallel."

💡 Tips

1. Phase sizing: Keep phases 1-4 hours max
2. Prompts: Keep under 30 lines for reliability
3. E2E last: Run E2E only after unit tests pass
4. Context recovery: Always include bd show in prompts
5. Parallel agents: Use SINGLE message with MULTIPLE Task tool calls
6. Grouping: Group tests by independence for maximum parallelism

📚 Related

• Ralph Wiggum Plugin
• Beads Issue Tracker
• Playwright Test MCP

Documentation

See SKILL.md for complete documentation.