# CloudTeam AI-Toolkit — End-to-End SDLC Productivity Framework > **Enterprise AI assistance powered by OpenSpec workflow - Build with specifications first, implementation second** [![Framework Status](https://img.shields.io/badge/Status-Production%20Ready-brightgreen)]() [![Platform](https://img.shields.io/badge/Platform-Claude%20Code-blue)]() [![Setup Time](https://img.shields.io/badge/Setup%20Time-%3C%205%20minutes-blue)]() [![Compliance](https://img.shields.io/badge/Compliance-ISO%2FCMMI%20Level%203-blue)]() --- ## What Is CloudTeam AI-Toolkit? CloudTeam AI-Toolkit is a **spec-driven development framework** that brings structure and compliance to AI-assisted software development. Instead of ad-hoc code changes, every modification follows a proven three-phase workflow: **Specification → Implementation → Validation**. ### Core Philosophy **Every change starts with a specification. Every implementation follows a spec. Every deployment is validated against its spec.** This approach ensures: - ✅ Clear requirements before coding - ✅ Automatic compliance validation (ISO/IEC 27001/27701/27002, CMMI Level 3) - ✅ Complete traceability from idea to production - ✅ Consistent quality across teams - ✅ No lost context or tribal knowledge ### What You Get - **OpenSpec Workflow** - Three commands to manage all changes - **10 Specialist AI Agents** - Backend, Frontend, QA, Architecture, Platform, Data, Business Analysis, Code Review, Debugging, Documentation - **12 Claude Skills** - Reusable capabilities for AI, frontend, backend, DevOps, authentication, media processing, and more - **16 Slash Commands** - Quick shortcuts for common workflows - **Compliance Built-In** - Automatic validation against industry standards - **MCP Server Integration** - Database adapters, Atlassian, Figma, Playwright, MongoDB, Context7 --- ## 🚀 Installation ### Global Installation Global installation allows you to use the toolkit across multiple projects without installing it in each one. #### Step 1: Install Globally **For macOS/Linux users:** The package will automatically set up user-prefix during installation (no `sudo` needed): ```bash npm install -g @cloudteam-dev/ai-toolkit ``` The installation will automatically: - Create `~/.npm-global` directory if it doesn't exist - Configure npm to use user-prefix - Add to your shell PATH (`.zshrc` or `.bashrc`) **Note:** After installation, restart your terminal or run `source ~/.zshrc` (or `source ~/.bashrc`) to load the new PATH. **For Windows users:** No setup needed - npm already uses user folder by default: ```bash npm install -g @cloudteam-dev/ai-toolkit ``` #### Step 2: Initialize in Your Project Navigate to your project directory and run: ```bash cd my-project # Initialize the toolkit (copies files to current directory) cteam-init # or cloudteam-ai-init ``` #### Step 3: Keep Updated When a new version is released on npm, update your toolkit: ```bash # Check for updates and sync project files cteam-update # or cloudteam-ai-update ``` This command will: - Check npm registry for the latest version - Update the global package if newer version exists - Re-sync project files with the latest toolkit version **IMPORTANT: After initialization, set up project documentation:** ```bash # Before starting any work, initialize documentation structure /docs:init ``` This command creates essential documentation files in the `docs/` folder: - `code-standards.md` - Coding standards and best practices - `codebase-summary.md` - High-level codebase overview - `project-overview-pdr.md` - Product Development Requirements - `system-architecture.md` - System architecture documentation Only after documentation initialization should you proceed with development. ### What Gets Installed The framework copies these directories to your project: ``` your-project/ ├── .claude/ # AI agents, commands, and skills ├── docs/ # Documentation templates and compliance standards ├── openspec/ # OpenSpec workflow directory ├── .mcp.json # MCP server configuration ├── CLAUDE.md # Project instructions for Claude ├── README.md # This file └── LICENSE # License information ``` --- ## 📖 Tutorial: OpenSpec Workflow The OpenSpec workflow has **three core commands** that guide you through every change: ### 1. `/openspec:proposal` - Create a Specification **What it does:** - Analyzes your requirements - Creates a structured specification document - Validates against ISO/CMMI compliance standards - Defines clear acceptance criteria - Generates implementation plan **When to use:** - Starting any new feature - Fixing bugs that require design decisions - Making architectural changes - Any work requiring planning **Example:** ``` /openspec:proposal Add OAuth2 authentication ``` **What you get:** A complete proposal in `openspec/changes/` with: - Change summary and rationale - Technical approach - Security considerations - Compliance checklist - Implementation steps - Testing strategy - Acceptance criteria **🚨 CRITICAL: Human Review Required** Before proceeding to implementation, **thoroughly review the entire proposal**. Validate **technical approach**, **security considerations**, **implementation steps**, and **acceptance criteria**. Request AI fixes for any **unreasonable points** or edit manually. **Never skip this step**—proceeding without validation risks implementing incorrect solutions. --- ### 2. `/openspec:apply` - Implement the Specification **What it does:** - Reads your approved specification - Implements features according to spec requirements - Tracks progress with todo items - Validates implementation against acceptance criteria - Updates proposal status **When to use:** - After stakeholders approve your proposal - When you're ready to implement the planned changes **Example:** ``` /openspec:apply oauth2-authentication ``` **What happens:** - Agent reads `openspec/changes/oauth2-authentication/` - Implements exactly what was specified - Creates todo list to track progress - Validates each step against requirements - Updates proposal with implementation notes --- ### 3. `/openspec:archive` - Archive After Deployment **⚠️ Before Archiving:** Verify all proposal tasks are complete and all acceptance criteria met. If incomplete, ask AI to continue implementation. Never archive partial work. **What it does:** - Marks proposal as deployed - Moves spec to archive with timestamp - Updates deployment records - Generates deployment summary **When to use:** - After successfully deploying to production - When the change is complete and validated **Example:** ``` /openspec:archive oauth2-authentication ``` **What happens:** - Moves `openspec/changes/oauth2-authentication/` to `openspec/changes/archive/YYMMDD-oauth2-authentication/` - Creates deployment record - Preserves full history for future reference --- ## 🤖 AI Agents CTEAM AI Kit includes **10 specialist agents** organized by department. Each agent has deep expertise in their domain and supports the OpenSpec workflow. ### Development Team **Backend Developer Agent** (`.claude/agents/dev/agent-backend-developer.md`) - API design, microservices, security, AI integration - Expertise: Node.js, Python, Go, databases, authentication - Shortcut: `/dev/as-dev-backend` **Frontend Developer Agent** (`.claude/agents/dev/agent-frontend-developer.md`) - UI/UX implementation, React components, accessibility, performance - Expertise: React, TypeScript, MUI, TanStack Router, responsive design - Shortcut: `/dev/as-dev-frontend` **Quality Engineer Agent** (`.claude/agents/dev/agent-quality-engineer.md`) - Test automation, compliance validation, security testing - Expertise: Unit tests, integration tests, E2E tests, test coverage - Shortcut: `/dev/as-dev-quality` ### Business & Analysis **Business Analyst Agent** (`.claude/agents/business/agent-business-analyst.md`) - Requirements engineering, stakeholder analysis, user stories - Creates specifications for OpenSpec proposals - Shortcut: `/ba/as-business-analyst` ### Infrastructure & Operations **Architecture Expert Agent** (`.claude/agents/infrastructure/agent-architecture-expert.md`) - System design, technical leadership, distributed systems - Designs technical architecture in proposals - Shortcut: `/infras/as-infra-architect` **Platform Engineer Agent** (`.claude/agents/infrastructure/agent-platform-engineer.md`) - Infrastructure as code, CI/CD, Kubernetes, cloud deployment - Expertise: Docker, Cloudflare, GCP, automation - Shortcut: `/infras/as-infra-platform` ### Data Engineering **Data Engineer Agent** (`.claude/agents/data/agent-data-engineer.md`) - Data pipelines, ETL, data warehouse, analytics - Implements data pipeline specifications - Shortcut: `/data/as-data-engineer` ### Specialized Agents **Code Reviewer Agent** (`.claude/agents/code-reviewer/`) - Code quality validation, security review, best practices - Validates code against specifications **Debugger Agent** (`.claude/agents/debugger/`) - Issue investigation, error analysis, diagnostics - Investigates issues during implementation **Docs Manager Agent** (`.claude/agents/docs/`) - Documentation generation, maintenance, updates - Maintains documentation throughout workflow ### How to Use Agents **Option 1: Use slash command shortcuts** ``` /dev/as-dev-backend Help me implement the payment gateway API ``` **Option 2: Direct invocation** ``` I need help with backend development for [task description] ``` The framework will automatically route to the appropriate specialist agent. --- ## 🎯 Claude Skills CTEAM AI Kit includes **12 reusable skills** that provide specialized capabilities: ### AI & Multimodal **AI Multimodal** (`.claude/skills/ai-multimodal/`) - Audio processing, video analysis, image generation, document extraction - Uses Google Gemini API for multimedia processing ### Development **Backend Development** (`.claude/skills/backend-development/`) - Node.js, Python, Go, APIs, databases, authentication, security **Frontend Development** (`.claude/skills/frontend-development/`) - React, TypeScript, MUI, TanStack Router, modern patterns **Frontend Design** (`.claude/skills/frontend-design/`) - UI/UX design, Figma integration, animations with anime.js **Aesthetic** (`.claude/skills/aesthetic/`) - Beautiful interface design, proven design principles, micro-interactions ### Infrastructure & DevOps **DevOps** (`.claude/skills/devops/`) - Cloudflare Workers, Docker, GCP deployment, edge computing **Better Auth** (`.claude/skills/better-auth/`) - Authentication framework integration, OAuth, 2FA, session management ### Utilities **Media Processing** (`.claude/skills/media-processing/`) - FFmpeg video/audio processing, ImageMagick image manipulation **Code Review** (`.claude/skills/code-review/`) - Quality assessment, security review, receiving feedback **Debugging** (`.claude/skills/debugging/`) - Systematic debugging, root cause tracing, verification **Docs Seeker** (`.claude/skills/docs-seeker/`) - Documentation search via llms.txt, library documentation lookup **Sequential Thinking** (`.claude/skills/sequential-thinking/`) - Structured problem-solving for complex tasks ### How Skills Work Skills are automatically activated when relevant to your task. You can also invoke them explicitly: ``` Use the ai-multimodal skill to analyze this video file ``` --- ## 💬 Slash Commands CTEAM AI Kit provides **16 slash commands** for quick access to common workflows: ### OpenSpec Workflow (Core) ```bash /openspec:proposal # Create new change specification /openspec:apply # Implement approved specification /openspec:archive # Archive deployed specification ``` ### Agent Shortcuts ```bash /dev/as-dev-backend # Backend Developer /dev/as-dev-frontend # Frontend Developer /dev/as-dev-quality # Quality Engineer /ba/as-business-analyst # Business Analyst /infras/as-infra-architect # Architecture Expert /infras/as-infra-platform # Platform Engineer /data/as-data-engineer # Data Engineer ``` ### Supporting Workflows **Git Operations:** ```bash /git/push # Auto-review, commit, and push /git/undo # Interactive undo /git/discard # Discard uncommitted changes /git/revert-last # Revert last commit ``` **Documentation:** ```bash /docs/init # Initialize project documentation /docs/update # Update existing documentation ``` **Debugging:** ```bash /debugger/debug # Launch debugging session ``` --- ## 📁 Project Structure ``` .claude/ ├── agents/ # 10 AI Agents │ ├── dev/ # Development (Backend, Frontend, QA) │ ├── business/ # Business Analysis │ ├── infrastructure/ # Architecture, Platform Engineering │ ├── data/ # Data Engineering │ ├── code-reviewer/ # Code Review │ ├── debugger/ # Debugging │ └── docs/ # Documentation ├── commands/ # 16 Slash Commands │ ├── openspec/ # OpenSpec workflow (CORE) │ ├── dev/ # Dev agent shortcuts │ ├── ba/ # Business analyst shortcut │ ├── infras/ # Infrastructure shortcuts │ ├── data/ # Data engineer shortcut │ ├── git/ # Git operations │ ├── docs/ # Documentation │ └── debugger/ # Debugging └── skills/ # 12 Reusable Skills docs/ ├── compliances/ # ISO/CMMI/PCI-DSS standards ├── brd/ # Business requirements ├── prd/ # Product requirements ├── epics/ # Epic-level documentation ├── tasks/ # Technical tasks └── user-stories/ # User stories openspec/ ├── changes/ # Active proposals │ └── archive/ # Deployed proposals └── specs/ # Specification templates ``` --- ## 🔒 Compliance & Standards CTEAM AI Kit automatically validates against: - ✅ **ISO/IEC 27001:2022** - Information Security Management - ✅ **ISO/IEC 27701:2019** - Privacy Information Management - ✅ **ISO/IEC 27002:2022** - Security Controls - ✅ **CMMI Level 3** - Defined Process Maturity - ✅ **PCI-DSS** - Payment Card Industry Data Security (when applicable) Compliance validation runs automatically in every OpenSpec proposal. --- ## 🔧 Requirements - **Claude Code** (for AI agents and MCP integration) - **Git** (for version control) - **Node.js** 18+ (optional, for dependencies) ### Optional (for MCP servers) - Docker (for Atlassian MCP server) - Database drivers (if using database adapter MCP servers) - Python 3.8+ (for document parsing scripts) --- ## 📚 Documentation - **CLAUDE.md** - Project instructions for Claude Code - **.claude/README.md** - Detailed agent documentation - **docs/compliances/** - ISO/CMMI/PCI-DSS standards - **openspec/** - OpenSpec proposals and archive --- ## 📞 Support - **Documentation**: Check docs/ directory first - **Contact**: phuc.phan@cloudteam.vn / ty.pham@cloudteam.vn - **Compliance Standards**: See docs/compliances/ for ISO/CMMI documentation --- ## 📜 Version History - **v1.0.0** *(November 2025)* - Initial release: Claude Code framework with department-based agent organization, MCP integration, OpenSpec workflow, 10 agents, 16 commands, 12 skills --- ## 📝 License ISC License - See LICENSE file for details --- ## 🔗 Quick Links - [OpenSpec Workflow](.claude/commands/openspec/) - Core spec-driven development commands - [AI Agents](.claude/agents/) - 10 specialist agents - [Claude Skills](.claude/skills/) - 12 reusable skills - [Slash Commands](.claude/commands/) - 16 commands total - [Compliance Standards](docs/compliances/) - ISO/IEC 27001/27701/27002, CMMI, PCI-DSS --- **Framework Version**: 1.0.0 **Last Updated**: November 29, 2025 **Platform**: Claude Code, Claude Desktop, Claude.ai **Methodology**: OpenSpec Workflow (Proposal → Apply → Archive)