{"id":32696477,"url":"https://github.com/michael-harris/claude-code-multi-agent-dev-system","last_synced_at":"2025-11-01T19:02:50.763Z","repository":{"id":321649864,"uuid":"1086649706","full_name":"michael-harris/claude-code-multi-agent-dev-system","owner":"michael-harris","description":"A multi-agent Claude Code workflow for Python \u0026 React","archived":false,"fork":false,"pushed_at":"2025-10-30T18:50:33.000Z","size":79,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-10-30T20:27:10.265Z","etag":null,"topics":["claude","claude-code","claude-code-agents"],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/michael-harris.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-10-30T17:50:50.000Z","updated_at":"2025-10-30T20:20:38.000Z","dependencies_parsed_at":"2025-10-30T20:37:49.789Z","dependency_job_id":null,"html_url":"https://github.com/michael-harris/claude-code-multi-agent-dev-system","commit_stats":null,"previous_names":["michael-harris/multi-agent-claude-workflow"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/michael-harris/claude-code-multi-agent-dev-system","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/michael-harris%2Fclaude-code-multi-agent-dev-system","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/michael-harris%2Fclaude-code-multi-agent-dev-system/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/michael-harris%2Fclaude-code-multi-agent-dev-system/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/michael-harris%2Fclaude-code-multi-agent-dev-system/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/michael-harris","download_url":"https://codeload.github.com/michael-harris/claude-code-multi-agent-dev-system/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/michael-harris%2Fclaude-code-multi-agent-dev-system/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":282188968,"owners_count":26628981,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-11-01T02:00:06.759Z","response_time":61,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["claude","claude-code","claude-code-agents"],"created_at":"2025-11-01T19:01:03.832Z","updated_at":"2025-11-01T19:02:50.755Z","avatar_url":"https://github.com/michael-harris.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# Multi-Agent Development System\n\nAn enterprise-grade Claude Code plugin providing **73 specialized AI agents** with hierarchical orchestration, T1/T2 cost optimization, and comprehensive support for **7 programming languages** across **10+ technology stacks**. Built for full-stack, multi-language, production-scale development with progress tracking and parallel development capabilities.\n\n---\n\n### 73 Specialized Agents\n\n**Planning Agents (3)**\n- **PRD Generator** - Interactive PRD creation with technology stack selection\n- **Task Graph Analyzer** - Breaks PRD into tasks with dependency analysis\n- **Sprint Planner** - Organizes tasks into executable sprints\n\n**Orchestration Agents (3)**\n- **Sprint Orchestrator** - Manages entire sprint execution (Opus)\n- **Task Orchestrator** - Coordinates single task with T1/T2 escalation (Sonnet)\n- **Requirements Validator** - Quality gate ensuring 100% criteria satisfaction (Opus)\n\n**Database Agents (15)** - *Expanded 3x for multi-language support*\n- **Database Designer** - Schema design and normalization (Opus)\n- **Database Developer Python T1/T2** - SQLAlchemy + Alembic (Haiku/Sonnet)\n- **Database Developer TypeScript T1/T2** - Prisma/TypeORM (Haiku/Sonnet)\n- **Database Developer Java T1/T2** - JPA/Hibernate + Flyway (Haiku/Sonnet)\n- **Database Developer C# T1/T2** - Entity Framework Core + migrations (Haiku/Sonnet)\n- **Database Developer Go T1/T2** - GORM + migrate (Haiku/Sonnet)\n- **Database Developer Ruby T1/T2** - ActiveRecord + migrations (Haiku/Sonnet)\n- **Database Developer PHP T1/T2** - Eloquent/Doctrine + migrations (Haiku/Sonnet)\n\n**Backend Agents (22)** - *Expanded 3x for enterprise stacks*\n- **API Designer** - RESTful API specifications with OpenAPI (Opus)\n- **API Developer Python T1/T2** - FastAPI/Django/Flask endpoints (Haiku/Sonnet)\n- **API Developer TypeScript T1/T2** - Express/NestJS/Fastify endpoints (Haiku/Sonnet)\n- **API Developer Java T1/T2** - Spring Boot/Micronaut endpoints (Haiku/Sonnet)\n- **API Developer C# T1/T2** - ASP.NET Core/Minimal API endpoints (Haiku/Sonnet)\n- **API Developer Go T1/T2** - Gin/Echo/Fiber endpoints (Haiku/Sonnet)\n- **API Developer Ruby T1/T2** - Rails/Sinatra/Grape endpoints (Haiku/Sonnet)\n- **API Developer PHP T1/T2** - Laravel/Symfony/Slim endpoints (Haiku/Sonnet)\n- **Backend Code Reviewers** - Python, TypeScript, Java, C#, Go, Ruby, PHP variants (Sonnet)\n\n**Frontend Agents (4)**\n- **Frontend Designer** - UI/UX component specifications (Opus)\n- **Frontend Developer T1/T2** - React/Vue/Next.js/Angular implementation (Haiku/Sonnet)\n- **Frontend Code Reviewer** - Accessibility, performance, best practices (Sonnet)\n\n**Python Agents (2)**\n- **Python Developer Generic T1/T2** - CLI tools, scripts, utilities (Haiku/Sonnet)\n\n**Quality Agents (10)** - *Expanded 3x for language-specific performance auditing*\n- **Test Writer** - Unit, integration, and e2e tests (Sonnet)\n- **Security Auditor** - OWASP Top 10 compliance (Opus)\n- **Documentation Coordinator** - Technical documentation (Sonnet)\n- **Performance Auditor Python** - Python-specific performance optimization (Sonnet)\n- **Performance Auditor TypeScript** - TypeScript/Node.js performance optimization (Sonnet)\n- **Performance Auditor Java** - Java/Spring Boot performance optimization (Sonnet)\n- **Performance Auditor C#** - .NET performance optimization (Sonnet)\n- **Performance Auditor Go** - Go performance optimization (Sonnet)\n- **Performance Auditor Ruby** - Ruby/Rails performance optimization (Sonnet)\n- **Performance Auditor PHP** - PHP/Laravel performance optimization (Sonnet)\n\n**Scripting Agents (4)** - *NEW*\n- **PowerShell Developer T1/T2** - Windows automation, Azure, DSC (Haiku/Sonnet)\n- **Shell Developer T1/T2** - Bash/Zsh automation scripts (Haiku/Sonnet)\n\n**DevOps Agents (4)** - *NEW*\n- **Docker Specialist** - Containerization, multi-stage builds, optimization (Sonnet)\n- **Kubernetes Specialist** - K8s manifests, Helm, operators (Sonnet)\n- **CI/CD Specialist** - GitHub Actions, GitLab CI, Jenkins (Sonnet)\n- **Terraform Specialist** - Infrastructure as Code for AWS/Azure/GCP (Sonnet)\n\n**Infrastructure Agents (2)** - *NEW*\n- **Configuration Manager T1/T2** - Config files, secrets, env management (Haiku/Sonnet)\n\n**Mobile Agents (4)** - *NEW*\n- **iOS Developer T1/T2** - SwiftUI, UIKit, native iOS development (Haiku/Sonnet)\n- **Android Developer T1/T2** - Kotlin, Jetpack Compose, native Android (Haiku/Sonnet)\n\n### Cost Optimization (T1/T2 System)\n\n**T1 Tier (Haiku)** - Cost-optimized first attempt\n- Handles 70-80% of implementation work\n- $0.001 per 1K tokens\n- Automatic escalation on failure\n\n**T2 Tier (Sonnet)** - Enhanced quality\n- Handles complex scenarios after T1 validation fails\n- 15-20% of work requiring deeper analysis\n- $0.003 per 1K tokens\n\n**Design Tier (Opus)** - High-value decisions\n- Schema design, API contracts, component architecture\n- Critical quality gates and security audits\n- 10% of work requiring architectural thinking\n- $0.015 per 1K tokens\n\n**Result: 60-70% cost savings vs all-Opus approach while maintaining quality**\n\n### Quality Gates\n\n- **100% Criteria Satisfaction** - Requirements validator enforces acceptance criteria\n- **Security First** - OWASP Top 10 audits on all implementations\n- **80%+ Test Coverage** - Enforced by test writer agent\n- **Code Review** - Language-specific reviewers for every implementation\n- **Iterative Refinement** - Max 5 iterations with T1→T2 escalation\n\n### Supported Languages \u0026 Frameworks\n\nThis system now provides enterprise-grade development coverage across **7 major programming languages** and **10+ development stacks**:\n\n| Language | Frameworks | ORM/Database | Testing | Code Review |\n|----------|-----------|--------------|---------|-------------|\n| **Python** | FastAPI, Django, Flask | SQLAlchemy, Alembic | pytest, unittest | ✅ Sonnet |\n| **TypeScript** | Express, NestJS, Fastify | Prisma, TypeORM | Jest, Vitest | ✅ Sonnet |\n| **Java** | Spring Boot, Micronaut | JPA/Hibernate, Flyway | JUnit, TestNG | ✅ Sonnet |\n| **C#** | ASP.NET Core, Minimal API | Entity Framework Core | xUnit, NUnit | ✅ Sonnet |\n| **Go** | Gin, Echo, Fiber | GORM, migrate | testing, testify | ✅ Sonnet |\n| **Ruby** | Rails, Sinatra, Grape | ActiveRecord | RSpec, minitest | ✅ Sonnet |\n| **PHP** | Laravel, Symfony, Slim | Eloquent, Doctrine | PHPUnit, Pest | ✅ Sonnet |\n\n**Frontend:**\n- React, Vue.js, Next.js, Angular (TypeScript)\n- Tailwind CSS, Material-UI, shadcn/ui\n- Vite, webpack, turbopack\n\n**Mobile:**\n- **iOS**: SwiftUI, UIKit, Combine\n- **Android**: Kotlin, Jetpack Compose, Coroutines\n\n**Database:**\n- PostgreSQL (primary), MySQL, SQLite, MongoDB\n\n**DevOps \u0026 Infrastructure:**\n- Docker, docker-compose, Kubernetes\n- GitHub Actions, GitLab CI, Jenkins\n- Terraform (AWS, Azure, GCP)\n- PowerShell, Bash/Zsh scripting\n- Configuration management (env, YAML, JSON, secrets)\n\n### Development Stacks Supported\n\n**Full-Stack Web Applications:**\n1. **MEAN/MERN Stack** - MongoDB/Express/Angular or React/Node.js\n2. **Python Stack** - FastAPI/Django + PostgreSQL + React\n3. **Java Enterprise** - Spring Boot + PostgreSQL + Angular\n4. **C# .NET Stack** - ASP.NET Core + SQL Server + React\n5. **Go Stack** - Gin/Echo + PostgreSQL + Vue\n6. **Ruby Stack** - Rails + PostgreSQL + React\n7. **PHP Stack** - Laravel + MySQL + Vue\n\n**Mobile Applications:**\n8. **iOS Native** - SwiftUI + CoreData + REST APIs\n9. **Android Native** - Kotlin + Room + Retrofit\n\n**DevOps \u0026 Infrastructure:**\n10. **Cloud Infrastructure** - Terraform + Docker + CI/CD\n11. **Microservices** - Any language + Docker + Kubernetes config\n\n**Scripting \u0026 Automation:**\n12. **System Administration** - Bash/Python scripts + Terraform\n13. **Data Processing** - Python + pandas + NumPy\n\n### New Capabilities\n\n**DevOps Integration**\n- Containerization with Docker specialist (multi-stage builds, optimization)\n- Kubernetes orchestration specialist (Helm charts, operators, manifests)\n- CI/CD pipeline specialist for GitHub Actions, GitLab CI, Jenkins\n- Infrastructure as Code with Terraform specialist for AWS, Azure, GCP\n- Automated deployment workflows\n\n**Mobile Development**\n- Native iOS development with SwiftUI and UIKit\n- Native Android development with Kotlin and Jetpack Compose\n- RESTful API integration and local data persistence\n- Platform-specific UI/UX best practices\n\n**Scripting \u0026 Automation**\n- PowerShell for Windows automation and Azure management\n- Shell scripting for Linux/Unix system automation (Bash/Zsh)\n- Configuration management for env files, YAML, JSON, secrets\n- Build automation and deployment scripts\n- Database migration and backup scripts\n\n**Enterprise Language Support**\n- Java enterprise applications with Spring Boot ecosystem\n- C# .NET Core applications for Windows/Linux/macOS\n- Go microservices with high-performance frameworks\n- Ruby web applications with Rails conventions\n- PHP applications with Laravel/Symfony frameworks\n\n## Installation\n\n### From GitHub (Recommended)\n\n```bash\n/plugin marketplace add https://github.com/michael-harris/claude-code-multi-agent-dev-system\n/plugin install multi-agent-dev-system\n```\n\n### From Local Path (Development)\n\n```bash\n# Clone the repository\ngit clone https://github.com/michael-harris/claude-code-multi-agent-dev-system.git\ncd claude-code-multi-agent-dev-system\n\n# Install locally\n./install-local.sh\n\n# OR manually\n/plugin marketplace add file:///absolute/path/to/claude-code-multi-agent-dev-system\n/plugin install multi-agent-dev-system\n```\n\n### Verify Installation\n\n```bash\n/plugin list\n# Should show: multi-agent-dev-system\n```\n\n## Quick Start\n\n### Full Workflow\n\n```bash\n# 1. Generate PRD\n/multi-agent:prd\n# Interactive interview to create comprehensive PRD\n# Output: docs/planning/PROJECT_PRD.yaml\n\n# 2. Create tasks and sprints\n/multi-agent:planning\n# Breaks PRD into tasks with dependencies\n# Organizes tasks into sprints\n# Output: docs/planning/tasks/TASK-*.yaml\n#         docs/sprints/SPRINT-*.yaml\n\n# 3. Execute sprint\n/multi-agent:sprint SPRINT-001\n# Automated execution with quality loops\n# T1 attempts first, escalates to T2 if needed\n# Requirements validator ensures 100% criteria met\n```\n\n### Individual Agent Usage\n\n```javascript\n// Design database schema\nTask(\n  subagent_type=\"multi-agent:database:designer\",\n  model=\"opus\",\n  prompt=\"Design schema for user authentication with roles and permissions\"\n)\n\n// Implement schema (T1 attempt)\nTask(\n  subagent_type=\"multi-agent:database:developer-python-t1\",\n  model=\"haiku\",\n  prompt=\"Implement the schema design from docs/api/database-schema.yaml using SQLAlchemy\"\n)\n\n// If T1 fails validation, escalate to T2\nTask(\n  subagent_type=\"multi-agent:database:developer-python-t2\",\n  model=\"sonnet\",\n  prompt=\"Fix the implementation issues identified by the validator\"\n)\n\n// Design API\nTask(\n  subagent_type=\"multi-agent:backend:api-designer\",\n  model=\"opus\",\n  prompt=\"Design REST API for user management\"\n)\n\n// Security audit\nTask(\n  subagent_type=\"multi-agent:quality:security-auditor\",\n  model=\"opus\",\n  prompt=\"Audit the authentication implementation for OWASP Top 10 vulnerabilities\"\n)\n```\n\n## Progress Tracking \u0026 Parallel Development\n\n### Progress Tracking (Resume Functionality)\n\nAll planning and execution workflows now include automatic progress tracking via state files:\n\n**State Files:**\n- Projects: `docs/planning/.project-state.yaml`\n- Features: `docs/planning/.feature-{id}-state.yaml`\n- Issues: `docs/planning/.issue-{id}-state.yaml`\n\n**Resume Capabilities:**\n- System tracks completion status of all tasks and sprints\n- Automatically skips completed work when resuming\n- Resume from any interruption point (system crash, manual stop, etc.)\n- State files provide detailed audit trail of development progress\n\n**Example:**\n```bash\n# Start sprint execution\n/multi-agent:sprint all\n\n# ... system interrupted at SPRINT-003 ...\n\n# Resume from where you left off\n/multi-agent:sprint all\n# System: \"Resuming from SPRINT-003 (SPRINT-001, SPRINT-002 already complete)\"\n```\n\n### Parallel Development Tracks\n\nEnable parallel development across independent task chains to dramatically reduce development time:\n\n**Single Track (Default):**\n```bash\n/multi-agent:planning           # Creates sprints: SPRINT-001, SPRINT-002, SPRINT-003\n/multi-agent:sprint all         # Sequential execution: ~128 hours\n```\n\n**Parallel Tracks (Fast):**\n```bash\n/multi-agent:planning 3         # Creates parallel tracks: SPRINT-001-01, SPRINT-001-02, SPRINT-001-03\n                                # System calculates max possible tracks from dependencies\n                                # Example output: \"Requested 3 tracks, max possible is 3. Using 3 tracks.\"\n\n# Execute all tracks in parallel (different terminals/sessions):\n/multi-agent:sprint all 01      # Terminal 1: Track 1 (~42 hours)\n/multi-agent:sprint all 02      # Terminal 2: Track 2 (~48 hours)\n/multi-agent:sprint all 03      # Terminal 3: Track 3 (~38 hours)\n\n# Result: ~48 hours (62% faster than sequential)\n```\n\n**How It Works:**\n1. **Task Graph Analyzer** calculates max parallel tracks from dependency analysis\n2. **Sprint Planner** distributes tasks across tracks using balanced bin-packing algorithm\n3. Each track contains independent task chains that can execute simultaneously\n4. If requested tracks \u003e max possible, system uses max and warns user\n\n**Benefits:**\n- 50-70% reduction in wall-clock development time\n- Ideal for projects with independent components (backend, frontend, infrastructure)\n- State tracking enables resumption for any track independently\n- Perfect for team collaboration (different tracks = different team members)\n\n### Development History Preservation\n\n**Sprint and task files are intentionally preserved:**\n- Provides complete development history and audit trail\n- Shows decision-making process and evolution of requirements\n- Valuable for debugging, rollback, and understanding context\n- Can be manually removed if desired, but recommended to keep\n- `.gitignore` can exclude state files (`.*.yaml`) if desired while keeping sprint definitions\n\n## Architecture\n\n### Hierarchical Orchestration\n\n```\nUser → /multi-agent:prd, /multi-agent:planning, /multi-agent:sprint commands\n   ↓\nSprint Orchestrator (Opus) - Manages entire sprint\n   ↓\nTask Orchestrator (Sonnet) - Coordinates single task with T1/T2 switching\n   ↓\nSpecialized Agents (T1/T2) - Implement with automatic escalation\n   ↓\nRequirements Validator (Opus) - Quality gate (100% criteria met)\n```\n\n### T1→T2 Escalation Flow\n\n```\nIteration 1: T1 agent (Haiku) attempts implementation\nIteration 2: T1 attempts fixes based on validator feedback\n  → If PASS: Task complete ✅\n  → If FAIL: Switch to T2 for iteration 3+\n\nIteration 3+: T2 agent (Sonnet) handles complexity\n\nMax Iterations: 5 before human intervention required\n```\n\n### Quality Gate System\n\nEvery task passes through:\n1. **Code Reviewers** - Language-specific quality checks\n2. **Security Auditor** - OWASP Top 10 compliance\n3. **Test Writer** - 80%+ coverage requirement\n4. **Requirements Validator** - Binary pass/fail on acceptance criteria\n\n**No task completes without 100% criteria satisfaction**\n\n## Model Distribution\n\n| Model | Count | Use Cases | Cost/1K Tokens |\n|-------|-------|-----------|----------------|\n| **Opus** | 6 | Design decisions, quality gates, security | $0.015 |\n| **Sonnet** | 45 | T2 developers, reviewers, orchestration, code review, performance auditing | $0.003 |\n| **Haiku** | 22 | T1 developers (first attempt across all languages) | $0.001 |\n\n**Total: 73 Agents**\n\n**Cost Optimization Logic:**\n1. T1 (Haiku) attempts implementation first (70-80% success rate)\n2. T2 (Sonnet) handles complex cases after T1 failure (15-20% of work)\n3. Opus only for critical design decisions and quality gates (10% of work)\n\n**Multi-Language Scaling:**\n- Each language gets T1/T2 developer pairs for database and backend work\n- Language-specific code reviewers ensure quality standards\n- Shared design agents (Opus) work across all languages\n- Same cost optimization applies to Java, C#, Go, Ruby, PHP as Python/TypeScript\n\n## Examples\n\n### Example 1: Complete Workflow\n\nSee [examples/complete-workflow-example.md](examples/complete-workflow-example.md) for a detailed walkthrough of building a task management application from PRD to deployment.\n\n### Example 2: Individual Agent Usage\n\nSee [examples/individual-agent-usage.md](examples/individual-agent-usage.md) for 10 targeted scenarios:\n- Quick database schema design\n- Implement database models from design\n- API design review\n- Implement API endpoint\n- Security audit\n- Generate tests for existing code\n- Create frontend component\n- Code review\n- Create CLI tool\n- Generate documentation\n\n### Example 3: Multi-Language Projects\n\nSee [examples/multi-language-examples.md](examples/multi-language-examples.md) for examples demonstrating multi-language support:\n- **Java Spring Boot** - Enterprise REST API with JPA\n- **C# .NET Core** - Minimal API with Entity Framework\n- **Go Microservice** - High-performance API with GORM\n- **Ruby on Rails** - Convention-based web application\n- **PHP Laravel** - Full-stack web application\n- **iOS Swift** - Native mobile app with SwiftUI\n- **Android Kotlin** - Native mobile app with Jetpack Compose\n- **DevOps Pipeline** - Complete CI/CD with Docker and Terraform\n\n### Example 4: Parallel Development Tracks\n\nSee [examples/parallel-tracks-example.md](examples/parallel-tracks-example.md) for a comprehensive walkthrough of using parallel development tracks to build an e-commerce platform with 50-70% reduction in wall-clock time.\n\n## Agent Reference\n\nComplete list of all 73 agents organized by category:\n\n### Planning (3 agents)\n- `planning:prd-generator` (Sonnet)\n- `planning:task-graph-analyzer` (Sonnet)\n- `planning:sprint-planner` (Sonnet)\n\n### Orchestration (3 agents)\n- `orchestration:sprint-orchestrator` (Opus)\n- `orchestration:task-orchestrator` (Sonnet)\n- `orchestration:requirements-validator` (Opus)\n\n### Database (15 agents)\n- `database:designer` (Opus)\n- `database:developer-python-t1` (Haiku)\n- `database:developer-python-t2` (Sonnet)\n- `database:developer-typescript-t1` (Haiku)\n- `database:developer-typescript-t2` (Sonnet)\n- `database:developer-java-t1` (Haiku)\n- `database:developer-java-t2` (Sonnet)\n- `database:developer-csharp-t1` (Haiku)\n- `database:developer-csharp-t2` (Sonnet)\n- `database:developer-go-t1` (Haiku)\n- `database:developer-go-t2` (Sonnet)\n- `database:developer-ruby-t1` (Haiku)\n- `database:developer-ruby-t2` (Sonnet)\n- `database:developer-php-t1` (Haiku)\n- `database:developer-php-t2` (Sonnet)\n\n### Backend (22 agents)\n- `backend:api-designer` (Opus)\n- `backend:api-developer-python-t1` (Haiku)\n- `backend:api-developer-python-t2` (Sonnet)\n- `backend:api-developer-typescript-t1` (Haiku)\n- `backend:api-developer-typescript-t2` (Sonnet)\n- `backend:api-developer-java-t1` (Haiku)\n- `backend:api-developer-java-t2` (Sonnet)\n- `backend:api-developer-csharp-t1` (Haiku)\n- `backend:api-developer-csharp-t2` (Sonnet)\n- `backend:api-developer-go-t1` (Haiku)\n- `backend:api-developer-go-t2` (Sonnet)\n- `backend:api-developer-ruby-t1` (Haiku)\n- `backend:api-developer-ruby-t2` (Sonnet)\n- `backend:api-developer-php-t1` (Haiku)\n- `backend:api-developer-php-t2` (Sonnet)\n- `backend:code-reviewer-python` (Sonnet)\n- `backend:code-reviewer-typescript` (Sonnet)\n- `backend:code-reviewer-java` (Sonnet)\n- `backend:code-reviewer-csharp` (Sonnet)\n- `backend:code-reviewer-go` (Sonnet)\n- `backend:code-reviewer-ruby` (Sonnet)\n- `backend:code-reviewer-php` (Sonnet)\n\n### Frontend (4 agents)\n- `frontend:designer` (Opus)\n- `frontend:developer-t1` (Haiku)\n- `frontend:developer-t2` (Sonnet)\n- `frontend:code-reviewer` (Sonnet)\n\n### Python (2 agents)\n- `python:developer-generic-t1` (Haiku)\n- `python:developer-generic-t2` (Sonnet)\n\n### Quality (10 agents)\n- `quality:test-writer` (Sonnet)\n- `quality:security-auditor` (Opus)\n- `quality:documentation-coordinator` (Sonnet)\n- `quality:performance-auditor-python` (Sonnet)\n- `quality:performance-auditor-typescript` (Sonnet)\n- `quality:performance-auditor-java` (Sonnet)\n- `quality:performance-auditor-csharp` (Sonnet)\n- `quality:performance-auditor-go` (Sonnet)\n- `quality:performance-auditor-ruby` (Sonnet)\n- `quality:performance-auditor-php` (Sonnet)\n\n### Scripting (4 agents) - NEW\n- `scripting:powershell-developer-t1` (Haiku)\n- `scripting:powershell-developer-t2` (Sonnet)\n- `scripting:shell-developer-t1` (Haiku)\n- `scripting:shell-developer-t2` (Sonnet)\n\n### DevOps (4 agents) - NEW\n- `devops:docker-specialist` (Sonnet)\n- `devops:kubernetes-specialist` (Sonnet)\n- `devops:cicd-specialist` (Sonnet)\n- `devops:terraform-specialist` (Sonnet)\n\n### Infrastructure (2 agents) - NEW\n- `infrastructure:configuration-manager-t1` (Haiku)\n- `infrastructure:configuration-manager-t2` (Sonnet)\n\n### Mobile (4 agents) - NEW\n- `mobile:ios-developer-t1` (Haiku)\n- `mobile:ios-developer-t2` (Sonnet)\n- `mobile:android-developer-t1` (Haiku)\n- `mobile:android-developer-t2` (Sonnet)\n\n## Repository Structure\n\n```\nclaude-code-multi-agent-dev-system/\n├── plugin.json              # Plugin manifest (73 agents, 6 commands)\n├── README.md                # This file\n├── LICENSE                  # MIT License\n├── install-local.sh         # Local installation script\n├── ENHANCEMENTS_V2.1.md     # Version 2.1 enhancement documentation\n├── agents/                  # 73 agent definitions\n│   ├── planning/           # PRD, task analysis, sprint planning (3)\n│   ├── orchestration/      # Sprint/task orchestration, validation (3)\n│   ├── database/           # Schema design + 7-language implementation (15)\n│   │   ├── designer.md\n│   │   ├── developer-python-t1.md\n│   │   ├── developer-python-t2.md\n│   │   ├── developer-typescript-t1.md\n│   │   ├── developer-typescript-t2.md\n│   │   ├── developer-java-t1.md\n│   │   ├── developer-java-t2.md\n│   │   ├── developer-csharp-t1.md\n│   │   ├── developer-csharp-t2.md\n│   │   ├── developer-go-t1.md\n│   │   ├── developer-go-t2.md\n│   │   ├── developer-ruby-t1.md\n│   │   ├── developer-ruby-t2.md\n│   │   ├── developer-php-t1.md\n│   │   └── developer-php-t2.md\n│   ├── backend/            # API design + 7-language implementation + review (22)\n│   │   ├── api-designer.md\n│   │   ├── api-developer-[language]-t1.md (7 languages)\n│   │   ├── api-developer-[language]-t2.md (7 languages)\n│   │   └── code-reviewer-[language].md (7 languages)\n│   ├── frontend/           # UI design + implementation + review (4)\n│   ├── python/             # Generic Python development (2)\n│   ├── quality/            # Testing, security, documentation (3)\n│   ├── scripting/          # PowerShell and Shell scripting (4) - NEW\n│   │   ├── powershell-developer-t1.md\n│   │   ├── powershell-developer-t2.md\n│   │   ├── shell-developer-t1.md\n│   │   └── shell-developer-t2.md\n│   ├── devops/             # Docker, Kubernetes, CI/CD, Terraform (4) - NEW\n│   │   ├── docker-specialist.md\n│   │   ├── kubernetes-specialist.md\n│   │   ├── cicd-specialist.md\n│   │   └── terraform-specialist.md\n│   ├── infrastructure/     # Configuration management (2) - NEW\n│   │   ├── configuration-manager-t1.md\n│   │   └── configuration-manager-t2.md\n│   ├── quality/            # Testing, security, documentation, performance (10)\n│   │   ├── test-writer.md\n│   │   ├── security-auditor.md\n│   │   ├── documentation-coordinator.md\n│   │   ├── performance-auditor-python.md\n│   │   ├── performance-auditor-typescript.md\n│   │   ├── performance-auditor-java.md\n│   │   ├── performance-auditor-csharp.md\n│   │   ├── performance-auditor-go.md\n│   │   ├── performance-auditor-ruby.md\n│   │   └── performance-auditor-php.md\n│   └── mobile/             # iOS and Android (4) - NEW\n│       ├── ios-developer-t1.md\n│       ├── ios-developer-t2.md\n│       ├── android-developer-t1.md\n│       └── android-developer-t2.md\n├── commands/               # 6 workflow commands\n│   ├── prd.md             # Generate PRD\n│   ├── planning.md        # Create tasks and sprints (with parallel tracks support)\n│   ├── sprint.md          # Execute single sprint\n│   ├── sprint-all.md      # Execute all sprints (with track filtering)\n│   ├── feature.md         # Complete feature workflow (PRD → Planning → Implementation)\n│   └── issue.md           # Complete issue resolution workflow\n├── examples/              # Usage examples\n│   ├── complete-workflow-example.md           # Full PRD-to-deployment walkthrough\n│   ├── individual-agent-usage.md             # 10 targeted scenarios\n│   ├── multi-language-examples.md            # Multi-language project examples\n│   └── parallel-tracks-example.md            # Parallel development demonstration\n└── docs/                  # Work directories and archives\n    ├── development/       # Archived development history\n    ├── adrs/              # Architecture Decision Records (generated)\n    ├── api/               # API documentation (generated)\n    ├── features/          # Feature specifications (generated)\n    ├── planning/          # PRDs and tasks (generated)\n    ├── reviews/           # Code reviews (generated)\n    └── sprints/           # Sprint definitions (generated)\n```\n\n## Cost Analysis\n\n### Plugin Development\n- **Manual Development (73 agents)**: 250-350 hours × $150/hour = $37,500-52,500\n- **AI-Assisted Build**: ~6 hours × $0.50/hour = $3.00\n- **Savings**: 99.99%\n\n### Using the Plugin (Per Project)\n\n**Cost per sprint remains similar despite 2.4x agent expansion:**\n- T1/T2 system ensures only relevant language agents are used\n- Most projects use 1-2 languages, not all 7\n- Shared design agents (Opus) work across all languages\n\n**Estimated project costs:**\n- **Small project (1 sprint, 1 language)**: ~$0.70-1.00\n- **Medium project (3 sprints, 2 languages)**: ~$6-10\n- **Large project (10 sprints, 3 languages)**: ~$30-40\n- **Enterprise polyglot (20 sprints, 5+ languages)**: ~$80-120\n\n**Multi-language example breakdown:**\n- Database design (Opus): $0.50\n- Database implementation Java (T1/T2): $0.30\n- API design (Opus): $0.60\n- API implementation Java (T1/T2): $0.40\n- Frontend implementation (T1/T2): $0.50\n- Testing + Security (Sonnet/Opus): $0.70\n- Code reviews (Sonnet): $0.30\n- **Total per sprint**: ~$3.30\n\n**Compared to:**\n- Human developers: 99%+ savings\n- All-Opus AI: 60-70% savings\n- Single-language AI systems: Same cost, 7x language coverage\n\n## Why This System?\n\n### Enterprise-Ready Multi-Language Support\n- **7 Programming Languages**: Python, TypeScript, Java, C#, Go, Ruby, PHP\n- **10+ Development Stacks**: From MEAN/MERN to Spring Boot to Rails\n- **Consistent Quality**: Same T1/T2 optimization and quality gates across all languages\n- **Language-Specific Expertise**: Dedicated code reviewers for each language\n\n### Complete Development Lifecycle\n- **Planning**: PRD generation, task breakdown, sprint planning\n- **Implementation**: Database, backend, frontend, mobile\n- **DevOps**: Containerization, CI/CD, infrastructure as code\n- **Quality**: Security audits, testing, code review, documentation\n\n### Production-Scale Features\n- **Mobile Development**: Native iOS (Swift) and Android (Kotlin)\n- **DevOps Automation**: Docker, GitHub Actions, GitLab CI, Jenkins\n- **Cloud Infrastructure**: Terraform for AWS, Azure, GCP\n- **Scripting**: Shell and Python automation scripts\n\n### Cost-Optimized Intelligence\n- **73 Agents**: Specialized for every language and task\n- **3-Tier System**: Haiku (fast/cheap) → Sonnet (balanced) → Opus (expert)\n- **Smart Escalation**: T1 handles 70-80% of work, T2 only when needed\n- **60-70% Cost Savings**: Compared to all-Opus approaches\n\n### Quality Without Compromise\n- **100% Criteria Satisfaction**: Requirements validator enforces every acceptance criteria\n- **Security First**: OWASP Top 10 audits on all implementations\n- **80%+ Test Coverage**: Automated test generation\n- **Language-Specific Review**: Code reviewers for Python, TypeScript, Java, C#, Go, Ruby, PHP\n\n## Use Cases\n\n**Greenfield Projects**\n- Build full-stack applications from scratch in any supported language\n- Generate PRD, design database and APIs, implement frontend/backend\n- Deploy with Docker and CI/CD pipelines\n\n**Multi-Language Enterprises**\n- Microservices architecture with different languages per service\n- Unified quality standards across Python, Java, Go, C# codebases\n- Consistent code review and security practices\n\n**Mobile + Backend**\n- Native iOS/Android apps with backend APIs\n- Choose optimal backend language (Python for ML, Go for performance, etc.)\n- Automated API client generation and integration\n\n**DevOps \u0026 Infrastructure**\n- Containerize existing applications with Docker\n- Create CI/CD pipelines for automated testing and deployment\n- Provision cloud infrastructure with Terraform\n\n**Legacy Modernization**\n- Migrate from one language/framework to another\n- Refactor monoliths to microservices\n- Add tests and security audits to existing code\n\n## Contributing\n\nContributions are welcome! Please feel free to submit issues or pull requests.\n\n**Potential Contributions:**\n- Additional language support (Rust, Scala, Kotlin backend, etc.)\n- New agent categories (testing frameworks, monitoring, etc.)\n- Language-specific example projects\n- Performance optimizations and best practices\n\n## License\n\nMIT License - See LICENSE file for details.\n\n## Development History\n\nThis plugin was developed through an iterative AI-assisted process. Development history and architectural decisions are archived in `docs/development/` for reference.\n\n**Evolution:**\n- **v1.0**: Initial release with 27 agents (Python/TypeScript only)\n- **v2.0**: Expanded to 66 agents with 7 languages, mobile, DevOps, infrastructure\n- **v2.1**: Expanded to 73 agents with language-specific performance auditing, progress tracking, and parallel development tracks\n\n## Support\n\n- **Issues**: https://github.com/michael-harris/claude-code-multi-agent-dev-system/issues\n- **Documentation**: See `examples/` directory\n- **Installation Help**: See installation section above\n- **Feature Requests**: Open an issue with the `enhancement` label\n\n## Frequently Asked Questions\n\n**Q: Do I need to use all 73 agents?**\nA: No. The system automatically selects relevant agents based on your project's language and requirements. Most projects use 10-15 agents.\n\n**Q: Can I mix languages in one project?**\nA: Yes. The system excels at polyglot projects. For example, Go backend + TypeScript frontend + Python ML services.\n\n**Q: What if my language isn't supported?**\nA: The generic Python agents can handle similar languages (like Ruby for scripts), or you can contribute new agents following the existing patterns.\n\n**Q: How does cost optimization work across languages?**\nA: All languages use the same T1/T2 system. Java, C#, Go, Ruby, PHP developers benefit from the same 60-70% cost savings as Python/TypeScript.\n\n**Q: Can I use this for existing projects?**\nA: Yes. You can use individual agents for specific tasks (add tests, create Dockerfile, generate API endpoints, etc.) without the full workflow.\n\n**Q: Do the mobile agents generate production-ready code?**\nA: Yes, with the same quality standards as web development: code review, security audits, testing, and iterative refinement with T1/T2 escalation.\n\n---\n\n**Built with Claude Code** - Demonstrating the power of enterprise-scale multi-agent AI development systems.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmichael-harris%2Fclaude-code-multi-agent-dev-system","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmichael-harris%2Fclaude-code-multi-agent-dev-system","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmichael-harris%2Fclaude-code-multi-agent-dev-system/lists"}