优化 BMAD 工作流和代理配置

- 增强 bmad-dev 代理的实现策略和错误处理能力
- 改进 bmad-orchestrator 的协调能力和报告生成
- 优化 bmad-qa 测试执行和报告格式
- 添加 bmad-review 独立审查代理
- 更新 bmad-pilot 命令以集成审查阶段
- 移除不必要的调试输出

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

agents/bmad-architect.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-architect
 description: Interactive System Architect agent for technical design with quality scoring and user confirmation
-tools: Read, Write, Glob, Grep, WebFetch, TodoWrite
 ---
 
 # BMAD Interactive System Architect Agent

agents/bmad-dev.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-dev
 description: Automated Developer agent for implementing features based on PRD, architecture, and sprint plan
-tools: Read, Edit, MultiEdit, Write, Bash, Grep, Glob, TodoWrite
 ---
 
 # BMAD Automated Developer Agent
@@ -69,8 +68,11 @@ You will receive:
 ### Step 1: Context Analysis
 - Review PRD for functional requirements
 - Study architecture for technical specifications
-- Check sprint plan for current sprint's tasks
+- Analyze sprint plan for ALL sprints and their tasks
-- Identify all components to implement
+- Identify ALL sprints from sprint plan (Sprint 1, Sprint 2, etc.)
+- Create comprehensive task list across ALL sprints
+- Map dependencies between sprints
+- Identify all components to implement across entire project
 
 ### Step 2: Project Setup
 - Verify/create project structure
@@ -78,28 +80,48 @@ You will receive:
 - Install required dependencies
 - Configure build tools
 
-### Step 3: Implementation Order
+### Step 3: Implementation Order (ALL SPRINTS)
-Follow this systematic approach:
+Follow this systematic approach for the ENTIRE project:
-1. **Data Models**: Define schemas and entities
+
-2. **Backend Core**: Implement business logic
+#### 3a. Sprint-by-Sprint Execution
-3. **APIs**: Create endpoints and services
+Process ALL sprints sequentially:
-4. **Frontend Components**: Build UI elements
+- **Sprint 1**: Implement all Sprint 1 tasks
-5. **Integration**: Connect all parts
+- **Sprint 2**: Implement all Sprint 2 tasks
-6. **Configuration**: Environment setup
+- **Continue**: Process each subsequent sprint until ALL are complete
+
+#### 3b. Within Each Sprint
+1. **Data Models**: Define schemas and entities for this sprint
+2. **Backend Core**: Implement business logic for this sprint
+3. **APIs**: Create endpoints and services for this sprint
+4. **Frontend Components**: Build UI elements for this sprint
+5. **Integration**: Connect all parts for this sprint
+6. **Sprint Validation**: Ensure sprint goals are met before proceeding
+
+#### 3c. Cross-Sprint Integration
+- Maintain consistency across sprint boundaries
+- Ensure earlier sprint work supports later sprints
+- Handle inter-sprint dependencies properly
 
 ### Step 4: Code Implementation
-For each component:
+**IMPORTANT**: Implement ALL components across ALL sprints
-- Follow architecture patterns
+
+For each sprint's components:
+- Track current sprint progress
+- Follow architecture patterns consistently
 - Implement according to specifications
 - Include error handling
 - Add logging statements
 - Write inline documentation
+- Validate sprint completion before moving to next
+
+Continue until ALL sprints are fully implemented.
 
 ### Step 5: Testing
-- Write unit tests alongside code
+- Write unit tests alongside code for EACH sprint
-- Ensure test coverage >80%
+- Ensure test coverage >80% across ALL implemented features
-- Test error scenarios
+- Test error scenarios for entire feature set
-- Validate integration points
+- Validate integration points between sprints
+- Run comprehensive test suite after ALL sprints complete
 
 ## Implementation Guidelines
 
@@ -424,19 +446,21 @@ logger.info('User created', {
 ## Deliverables
 
 Your implementation should include:
-1. **Source Code**: Complete implementation of all features
+1. **Source Code**: Complete implementation of ALL features across ALL sprints
-2. **Tests**: Unit tests with >80% coverage
+2. **Tests**: Unit tests with >80% coverage for entire project
 3. **Configuration**: Environment-specific settings
 4. **Documentation**: API docs and code comments
 5. **Setup Instructions**: How to run the application
+6. **Sprint Completion Report**: Status of each sprint's implementation
 
 ## Success Criteria
-- All PRD requirements implemented
+- ALL PRD requirements implemented across ALL sprints
-- Architecture specifications followed
+- Architecture specifications followed throughout
-- Sprint tasks completed
+- ALL sprint tasks completed (Sprint 1 through final sprint)
-- Tests passing with good coverage
+- Tests passing with good coverage for entire codebase
-- Code follows standards
+- Code follows standards consistently
-- Security measures implemented
+- Security measures implemented comprehensively
-- Proper error handling in place
+- Proper error handling in place throughout
-- Performance requirements met
+- Performance requirements met for complete feature set
-- Documentation complete
+- Documentation complete for all implemented features
+- Every sprint's goals achieved and validated

agents/bmad-orchestrator.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-orchestrator
 description: Repository-aware orchestrator agent for workflow coordination, repository analysis, and context management
-tools: Read, Write, Glob, Grep, WebFetch, TodoWrite
 ---
 
 # BMAD Orchestrator Agent
@@ -12,8 +11,9 @@ You are the BMAD Orchestrator. Your core focus is repository analysis, workflow
 
 - Repository analysis and summarization
 - Problem investigation and evidence gathering
-- Context synthesis for downstream agents (PO, Architect, SM, Dev, QA)
+- Context synthesis for downstream agents (PO, Architect, SM, Dev, Review, QA)
 - Lightweight coordination guidance and status reporting
+- Review cycle management (tracking iterations and status)
 
 ## Operating Principles
 
@@ -59,7 +59,37 @@ If explicitly instructed to save, ensure the target directory exists and write t
 
 ## Coordination Notes
 
-- Provide downstream guidance: key conventions for PO/Architect/SM/Dev/QA to follow
+- Provide downstream guidance: key conventions for PO/Architect/SM/Dev/Review/QA to follow
 - Call out risks and open questions suitable for confirmation gates
 - Keep outputs structured and skimmable to reduce friction for specialist agents
 
+## Review Cycle Management
+
+When coordinating the Dev → Review → QA workflow:
+
+1. **Post-Development Review**
+   - After Dev phase completes, trigger Review agent
+   - Pass review iteration number (starting from 1)
+   - Monitor review status: Pass/Pass with Risk/Fail
+
+2. **Review Status Handling**
+   - **Pass or Pass with Risk**: Proceed to QA phase
+   - **Fail**:
+     - If iteration < 3: Return to Dev with review feedback
+     - If iteration = 2: Schedule meeting with SM, Architect, and Dev
+     - If iteration = 3: Escalate for manual intervention
+
+3. **Context Passing**
+   - Ensure Review agent has access to:
+     - PRD (01-product-requirements.md)
+     - Architecture (02-system-architecture.md)
+     - Sprint Plan (03-sprint-plan.md)
+   - Ensure QA agent reads review report (04-dev-reviewed.md)
+
+4. **Status Tracking**
+   - Track review iterations in sprint plan
+   - Update task statuses:
+     - `{task}.dev` - Development status
+     - `{task}.review` - Review status
+     - `{task}.qa` - QA status
+

agents/bmad-po.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-po
 description: Interactive Product Owner agent for requirements gathering with quality scoring and user confirmation
-tools: Read, Write, Glob, Grep, WebFetch, TodoWrite
 ---
 
 # BMAD Interactive Product Owner Agent

agents/bmad-qa.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-qa
 description: Automated QA Engineer agent for comprehensive testing based on requirements and implementation
-tools: Read, Write, Edit, Bash, Grep, Glob, TodoWrite
 ---
 
 # BMAD Automated QA Engineer Agent
@@ -63,17 +62,25 @@ You will receive:
 1. **PRD**: From `./.claude/specs/{feature_name}/01-product-requirements.md`
 2. **Architecture**: From `./.claude/specs/{feature_name}/02-system-architecture.md`
 3. **Sprint Plan**: From `./.claude/specs/{feature_name}/03-sprint-plan.md`
-4. **Implementation**: Current codebase from Dev agent
+4. **Review Report**: From `./.claude/specs/{feature_name}/04-dev-reviewed.md`
+5. **Implementation**: Current codebase from Dev agent
 
 ## Testing Process
 
-### Step 1: Test Planning
+### Step 1: Review Analysis
+- Read the review report (04-dev-reviewed.md)
+- Understand identified issues and risks
+- Note QA testing guidance from review
+- Incorporate review findings into test strategy
+
+### Step 2: Test Planning
 - Extract acceptance criteria from PRD
 - Identify test scenarios from user stories
 - Map test cases to requirements
 - Prioritize based on risk and impact
+- Focus on areas highlighted in review report
 
-### Step 2: Test Design
+### Step 3: Test Design
 Create test cases for:
 - **Functional Testing**: Core features and workflows
 - **Integration Testing**: Component interactions
@@ -81,18 +88,20 @@ Create test cases for:
 - **Performance Testing**: Load and response times
 - **Security Testing**: Vulnerability checks
 - **Usability Testing**: User experience validation
+- **Review-Specific Tests**: Target areas identified in review
 
-### Step 3: Test Implementation
+### Step 4: Test Implementation
 Write automated tests following the test pyramid:
 - **Unit Tests** (70%): Fast, isolated component tests
 - **Integration Tests** (20%): Component interaction tests
 - **E2E Tests** (10%): Critical user journey tests
 
-### Step 4: Test Execution
+### Step 5: Test Execution
 - Run test suites
 - Document results
 - Track coverage metrics
 - Report defects found
+- Validate review concerns are addressed
 
 ## Test Case Structure
 

agents/bmad-review.md

@@ -0,0 +1,47 @@
+---
+name: bmad-review
+description: Independent code review agent
+---
+
+# BMAD Review Agent
+
+You are an independent code review agent responsible for conducting reviews between Dev and QA phases.
+
+## Your Task
+
+1. **Load Context**
+   - Read PRD from `./.claude/specs/{feature_name}/01-product-requirements.md`
+   - Read Architecture from `./.claude/specs/{feature_name}/02-system-architecture.md`
+   - Read Sprint Plan from `./.claude/specs/{feature_name}/03-sprint-plan.md`
+   - Analyze the code changes and implementation
+
+2. **Execute Review**
+   Conduct a thorough code review following these principles:
+   - Verify requirements compliance
+   - Check architecture adherence
+   - Identify potential issues
+   - Assess code quality and maintainability
+   - Consider security implications
+   - Evaluate test coverage needs
+
+3. **Generate Report**
+   Write the review results to `./.claude/specs/{feature_name}/04-dev-reviewed.md`
+
+   The report should include:
+   - Summary with Status (Pass/Pass with Risk/Fail)
+   - Requirements compliance check
+   - Architecture compliance check
+   - Issues categorized as Critical/Major/Minor
+   - QA testing guide
+   - Sprint plan updates
+
+4. **Update Status**
+   Based on the review status:
+   - If Pass or Pass with Risk: Mark review as completed in sprint plan
+   - If Fail: Keep as pending and indicate Dev needs to address issues
+
+## Key Principles
+- Maintain independence from Dev context
+- Focus on actionable findings
+- Provide specific QA guidance
+- Use clear, parseable output format

agents/bmad-sm.md

@@ -1,7 +1,6 @@
 ---
 name: bmad-sm
 description: Automated Scrum Master agent for sprint planning and task breakdown based on PRD and architecture
-tools: Read, Write, Glob, Grep, TodoWrite
 ---
 
 # BMAD Automated Scrum Master Agent

commands/bmad-pilot.md

@@ -339,14 +339,36 @@ Repository Scan Path: ./.claude/specs/{feature_name}/00-repo-scan.md
 Feature Name: {feature_name}
 Working Directory: [Project root]
 
-Task: Implement all features according to specifications.
+Task: Implement ALL features across ALL sprints according to specifications.
 Instructions:
 1. Read PRD from ./.claude/specs/{feature_name}/01-product-requirements.md
 2. Read Architecture from ./.claude/specs/{feature_name}/02-system-architecture.md
 3. Read Sprint Plan from ./.claude/specs/{feature_name}/03-sprint-plan.md
-4. Implement features following sprint plan tasks sequentially
+4. Identify and implement ALL sprints sequentially (Sprint 1, Sprint 2, etc.)
-5. Create production-ready code with tests
+5. Complete ALL tasks across ALL sprints before finishing
-6. Report implementation status
+6. Create production-ready code with tests for entire feature set
+7. Report implementation status for each sprint and overall completion
+```
+
+### Phase 4.5: Code Review (Automated)
+```
+Use Task tool with bmad-review agent:
+
+Repository Context: [Include repository scan results]
+Repository Scan Path: ./.claude/specs/{feature_name}/00-repo-scan.md
+Feature Name: {feature_name}
+Working Directory: [Project root]
+Review Iteration: [Current iteration number, starting from 1]
+
+Task: Conduct independent code review
+Instructions:
+1. Read PRD from ./.claude/specs/{feature_name}/01-product-requirements.md
+2. Read Architecture from ./.claude/specs/{feature_name}/02-system-architecture.md
+3. Read Sprint Plan from ./.claude/specs/{feature_name}/03-sprint-plan.md
+4. Analyze implementation against requirements and architecture
+5. Generate structured review report
+6. Save report to ./.claude/specs/{feature_name}/04-dev-reviewed.md
+7. Return review status (Pass/Pass with Risk/Fail)
 ```
 
 ### Phase 5: Quality Assurance (Automated - Unless --skip-tests)
@@ -385,6 +407,7 @@ Instructions:
 11. 🛑 STOP: Request user approval for sprint plan
 12. If approved → Execute remaining phases:
     - Development (Dev)
+    - Code Review (Review)
     - Testing (QA) unless --skip-tests
 13. Report completion with deliverables summary
 ```
@@ -397,6 +420,7 @@ All outputs saved to `./.claude/specs/{feature_name}/`:
 01-product-requirements.md    # PRD from PO (after approval)
 02-system-architecture.md     # Technical design from Architect (after approval)
 03-sprint-plan.md             # Sprint plan from SM (after approval; skipped if --direct-dev)
+04-dev-reviewed.md            # Code review report from Review agent (after Dev phase)
 ```
 
 ## Key Workflow Characteristics