auto-dev-pipeline（自动开发管道）

Auto Dev Pipeline - One-Person Company Development Automation

Overview

The Auto Dev Pipeline is a complete automated development system that transforms natural language app ideas into fully tested iOS applications. It orchestrates three specialized skills to create a seamless, hands-off development process:

1. PRD Generation (prd-skill): Requirements → Structured PRD
2. Development (dev-skill): PRD → SwiftUI iOS Code
3. Quality Assurance (qa-skill): Code → Test Cases & Validation

Pipeline Architecture

1. Trigger Mechanism

The pipeline is triggered by natural language app ideas:

• "做一个待办事项App"
• "开发一个健身追踪应用"
• "创建一个社交网络应用"

2. Automated Coordination

The pipeline uses OpenClaw's session management to automatically:

1. Spawn prd-skill sub-agent with user requirements
2. Monitor PRD completion and trigger dev-skill
3. Monitor code generation and trigger qa-skill
4. Collect final outputs and provide summary

3. Data Flow

User Input → prd-skill → PRD Document → dev-skill → SwiftUI Project → qa-skill → Test Suite

Complete Workflow

Phase 1: Requirements Analysis (prd-skill)

Input: Natural language app description Process:

1. Parse and analyze requirements
2. Generate structured PRD with:

- Product overview and target audience - Functional requirements with priorities - User flows and screen specifications - Technical requirements and constraints

1. Save PRD to output/prd/[timestamp]-[app-name].md

Auto-Trigger: Upon PRD completion, spawn dev-skill with PRD as input

Phase 2: Development Implementation (dev-skill)

Input: PRD document from Phase 1 Process:

1. Analyze PRD for technical requirements
2. Generate complete SwiftUI project with:

- MVVM architecture - Data models and services - UI components and navigation - Business logic implementation

1. Create Xcode project in output/dev/[app-name]/

Auto-Trigger: Upon code generation, spawn qa-skill with project as input

Phase 3: Quality Assurance (qa-skill)

Input: SwiftUI project from Phase 2 Process:

1. Analyze code structure and requirements
2. Generate comprehensive test suite:

- Unit tests for business logic - UI tests for user flows - Integration tests for data flow

1. Create test documentation and quality report
2. Save to output/qa/[app-name]-tests/

Completion: Pipeline ends with final summary and deliverables

Session Management

Sub-Agent Spawning

# Example coordination logic
def trigger_pipeline(user_requirements):
    # Step 1: Spawn PRD skill
    prd_session = sessions_spawn(
        task=f"Generate PRD for: {user_requirements}",
        runtime="subagent",
        agentId="prd-skill"
    )
    
    # Step 2: Monitor and trigger dev skill
    wait_for_completion(prd_session)
    prd_output = read_prd_output()
    
    dev_session = sessions_spawn(
        task=f"Develop iOS app from PRD: {prd_output}",
        runtime="subagent", 
        agentId="dev-skill"
    )
    
    # Step 3: Monitor and trigger QA skill
    wait_for_completion(dev_session)
    code_output = read_code_output()
    
    qa_session = sessions_spawn(
        task=f"Generate tests for: {code_output}",
        runtime="subagent",
        agentId="qa-skill"
    )
    
    # Step 4: Collect results
    wait_for_completion(qa_session)
    return compile_final_report()

Error Handling

• PRD Generation Failures: Retry with clarified requirements
• Code Generation Errors: Fallback to simpler implementation
• Test Generation Issues: Provide manual test guidelines
• Session Timeouts: Resume from last successful checkpoint

Output Structure

output/
├── prd/
│   ├── 20240319-1430-todo-app.md
│   └── 20240319-1500-fitness-tracker.md
├── dev/
│   ├── TodoApp/
│   │   ├── TodoApp.xcodeproj
│   │   ├── Sources/
│   │   └── README.md
│   └── FitnessTracker/
│       ├── FitnessTracker.xcodeproj
│       ├── Sources/
│       └── README.md
└── qa/
    ├── TodoApp-tests/
    │   ├── UnitTests/
    │   ├── UITests/
    │   └── TestReport.md
    └── FitnessTracker-tests/
        ├── UnitTests/
        ├── UITests/
        └── TestReport.md

Example: Complete Pipeline Execution

User Input

"做一个待办事项App，支持分类、提醒和分享功能"

Pipeline Execution

1. Phase 1 (PRD): 2 minutes

- Output: output/prd/20240319-1430-todo-app.md - Contains: 5 sections, 15 features, technical specs

1. Phase 2 (Development): 5 minutes

- Output: output/dev/TodoApp/ (Xcode project) - Contains: 12 Swift files, Core Data model, UI components

1. Phase 3 (QA): 3 minutes

- Output: output/qa/TodoApp-tests/ (Test suite) - Contains: 28 test cases, test plan, quality report

Final Delivery

• Total Time: 10 minutes
• Code Coverage: 85%
• Features Implemented: 12/15 (P0+P1)
• Test Cases: 28 automated tests
• Ready for: Xcode build and deployment

Configuration Options

Model Selection

pipeline:
  prd_model: "deepseekchat"  # For requirements analysis
  dev_model: "deepseekchat"  # For code generation  
  qa_model: "deepseekchat"   # For test generation

Output Customization

output:
  directory: "./auto-dev-output"
  keep_intermediate: true
  generate_readme: true
  include_build_instructions: true

Quality Settings

quality:
  min_code_coverage: 70
  require_ui_tests: true
  accessibility_check: true
  performance_benchmarks: true

Best Practices

For Users

1. Be Specific: Provide clear app descriptions
2. Set Expectations: Understand MVP vs full feature set
3. Review Outputs: Check PRD before development starts
4. Provide Feedback: Help improve pipeline accuracy

For Pipeline Maintenance

1. Monitor Performance: Track execution times and success rates
2. Update Skills: Keep prd/dev/qa skills current with best practices
3. Collect Metrics: Measure code quality and user satisfaction
4. Iterate Improvements: Continuously enhance automation logic

Troubleshooting

Common Issues

1. Vague Requirements: Pipeline asks for clarification
2. Complex Features: May require manual intervention
3. Technical Constraints: iOS limitations are documented
4. Timeouts: Pipeline resumes from last checkpoint

Resolution Steps

1. Check session logs for error details
2. Review intermediate outputs
3. Adjust requirements and retry
4. Contact pipeline maintainer for complex issues

Future Enhancements

Planned Features

1. Deployment Automation: App Store Connect integration
2. CI/CD Pipeline: GitHub Actions automation
3. Design Generation: Figma mockup creation
4. Documentation: User manuals and API docs
5. Monitoring: App analytics and crash reporting

Integration Opportunities

1. App Store: Automated submission and review
2. Backend Services: Firebase/CloudKit integration
3. Analytics: Mixpanel/Amplitude setup
4. Marketing: App store optimization tools