Claude Code Workflow (CCW)

Languages: English | 中文

A sophisticated multi-agent automation workflow framework that transforms complex software development tasks from conceptualization to implementation review into manageable, trackable, AI-orchestrated processes.

🎉 v1.0 Release: Complete Gemini CLI integration with template system, dynamic template discovery, streamlined documentation, and intelligent auto-selection capabilities. See CHANGELOG.md for details.

🏗️ Architecture Overview

Claude Code Workflow (CCW) is built on three foundational pillars:

JSON-Only Data Model

  • Single Source of Truth: All task states stored exclusively in .task/impl-*.json files
  • Dynamic Document Generation: Markdown files generated on-demand as read-only views
  • Zero Synchronization: Eliminates data consistency issues and sync complexity
  • Performance: Direct JSON operations with <1ms query times

Marker File Session Management

  • Ultra-Fast Operations: Session switching through atomic file operations (.workflow/.active-[session])
  • Self-Healing: Automatic detection and resolution of session conflicts
  • Visual Management: ls .workflow/.active-* shows current active session
  • Scalability: Supports hundreds of concurrent sessions without performance degradation

Progressive Complexity

CCW intelligently adapts its file structure and workflow processes based on unified task-count thresholds:

  • Simple workflows (<5 tasks): Minimal structure, single-level hierarchy
  • Medium workflows (5-15 tasks): Enhanced structure with progress tracking
  • Complex workflows (>15 tasks): Complete document suite with 3-level task decomposition

🚀 Core Features

Multi-Agent System

  • Conceptual Planning Agent: Multi-perspective brainstorming and concept planning
  • Action Planning Agent: Converts high-level concepts into executable implementation plans
  • Code Developer: Implements code based on plans
  • Code Review Agent: Reviews code quality and compliance
  • Memory Gemini Bridge: Intelligent CLAUDE.md documentation system with context-aware updates

Gemini CLI Integration (v1.0)

  • Dynamic Template Discovery: Automatically detects and loads templates from ~/.claude/prompt-templates/
  • Intelligent Auto-Selection: Matches user input against template keywords and descriptions
  • Template System: Bug-fix, planning, and custom analysis templates
  • Streamlined Commands: Consolidated documentation with 500+ lines reduced

Workflow Session Management

  • Create, pause, resume, list, and switch workflow sessions
  • Automatic initialization of required file and directory structures
  • Hierarchical workflow filesystem (.workflow/WFS-[topic-slug]/)

Intelligent Context Generation

  • Dynamic context construction based on technology stack detection
  • Project structure analysis and domain keyword extraction
  • Optimized file targeting for Gemini CLI integration

Dynamic Change Management

  • Issue tracking and integration (/workflow:issue)
  • Automatic re-planning capabilities (/task:replan)
  • Seamless adaptation to changing requirements

📁 Directory Structure

.claude/
├── agents/                 # AI agent definitions and behaviors
├── commands/              # CLI command implementations  
├── output-styles/         # Output formatting templates
├── planning-templates/    # Role-specific planning approaches
├── prompt-templates/      # AI interaction templates
├── scripts/              # Automation scripts
├── tech-stack-templates/ # Technology-specific templates
├── workflows/            # Core system architecture (v2.0)
│   ├── system-architecture.md     # 🆕 Unified architecture overview
│   ├── data-model.md              # 🆕 JSON-only task management spec
│   ├── complexity-rules.md        # 🆕 Unified complexity standards
│   ├── session-management-principles.md # Marker file session system
│   ├── file-structure-standards.md     # Progressive structure definitions
│   └── [gemini-*.md]              # Gemini CLI integration templates
└── settings.local.json   # Local configuration

.workflow/                 # 🆕 Session workspace (auto-generated)
├── .active-[session-name] # 🆕 Active session marker file
└── WFS-[topic-slug]/      # Individual session directories
    ├── workflow-session.json      # Session metadata
    ├── .task/impl-*.json          # 🆕 JSON-only task definitions
    ├── IMPL_PLAN.md               # Generated planning document
    └── .summaries/                # Generated completion summaries

🚀 Quick Start

Prerequisites

Install and configure Gemini CLI for optimal workflow integration.

Installation

One-liner installation:

Invoke-Expression (Invoke-WebRequest -Uri "https://raw.githubusercontent.com/catlog22/Claude-Code-Workflow/main/install-remote.ps1" -UseBasicParsing).Content

Verify installation:

/workflow:session list

Essential Configuration

For Gemini CLI integration, configure your settings.json file:

{
  "contextFileName": "CLAUDE.md"
}

⚠️ Important: Set "contextFileName": "CLAUDE.md" in your Gemini CLI settings.json to ensure proper integration with CCW's intelligent documentation system. This can be set in your user settings (~/.gemini/settings.json) or project settings (.gemini/settings.json).

📚 Complete Command Reference

Core Commands

Command Syntax Description
/enhance-prompt /enhance-prompt <input> Enhance and structure user inputs with technical context
/gemini:analyze /gemini:analyze <inquiry> [--all-files] [--save-session] Direct codebase analysis and investigation
/gemini:chat /gemini:chat <inquiry> [--all-files] [--save-session] Simple direct interaction with Gemini CLI without templates
/gemini:execute /gemini:execute <task-id|description> [--yolo] [--debug] Intelligent executor with automatic file context inference
/gemini:mode:auto /gemini:mode:auto "<description>" 🆕 Auto-select and execute appropriate template based on user input analysis
/gemini:mode:bug-index /gemini:mode:bug-index <bug-description> Bug analysis using specialized diagnostic template
/gemini:mode:plan /gemini:mode:plan <planning-topic> Project planning using specialized architecture template
/update-memory /update-memory [related|full] Intelligent CLAUDE.md documentation system with context-aware updates
/update-memory-full /update-memory-full 🆕 Complete project-wide CLAUDE.md documentation update with depth-parallel execution
/update-memory-related /update-memory-related 🆕 Context-aware documentation updates for modules affected by recent changes

Workflow Management

Command Syntax Description
/workflow:session:* /workflow:session:start|pause|resume|list|switch|status "task" Session lifecycle management with complexity adaptation
/workflow:brainstorm /workflow:brainstorm <topic> [--perspectives=role1,role2] Multi-agent conceptual planning from different expert perspectives
/workflow:plan [--from-brainstorming] [--skip-brainstorming] Convert concepts to executable implementation plans
/workflow:plan-deep <topic> [--complexity=high] [--depth=3] Deep architectural planning with comprehensive analysis
/workflow:execute [--type=simple|medium|complex] [--auto-create-tasks] Enter implementation phase with complexity-based organization
/workflow:review [--auto-fix] Final quality assurance with automated testing and validation
/workflow:issue:* create|list|update|close [options] 🆕 Dynamic issue and change request management
/context [task-id|--filter] [--analyze] [--format=tree|list|json] Unified task and workflow context with automatic data consistency

Task Execution

Command Syntax Description
/task:create "<title>" [--type=type] [--priority=level] Create hierarchical implementation tasks with auto-generated IDs
/task:breakdown <task-id> [--strategy=auto|interactive] [--depth=1-3] Intelligent task decomposition into manageable sub-tasks
/task:execute <task-id> [--mode=auto|guided] [--agent=type] Execute tasks with automatic agent selection
/task:replan [task-id|--all] [--reason] [--strategy=adjust|rebuild] Dynamic task re-planning for changing requirements

🎯 Usage Workflows

Complex Feature Development

# 1. Start sophisticated workflow with full documentation
/workflow:session:start "Implement OAuth2 authentication system"

# 2. Multi-perspective brainstorming
/workflow:brainstorm "OAuth2 architecture design" --perspectives=system-architect,security-expert,data-architect

# 3. Create detailed implementation plan
/workflow:plan --from-brainstorming

# 4. Break down into manageable tasks
/task:create "Backend API development"
/task:breakdown IMPL-1 --strategy=auto

# 5. Execute with intelligent automation
/gemini:execute IMPL-1.1 --yolo
/gemini:execute IMPL-1.2 --yolo

# 6. Handle dynamic changes and issues
/workflow:issue:create "Add social login support"
/workflow:issue:list
/workflow:issue:update 1 --status=in-progress

# 7. Monitor and review
/context --format=hierarchy
/workflow:review --auto-fix

Quick Bug Fix

# 1. Lightweight session for simple tasks
/workflow:session:start "Fix login button alignment"

# 2. Direct analysis and implementation
/gemini:analyze "Analyze login button CSS issues in @{src/components/Login.js}"

# 3. Create and execute single task
/task:create "Apply CSS fix to login button"
/task:execute IMPL-1 --mode=auto

# 4. Quick review
/workflow:review

Smart Template Auto-Selection (v1.0)

# 1. Automatic template selection based on keywords
/gemini:mode:auto "React component not rendering after state update"
# → Auto-selects bug-fix template

# 2. Planning template for architecture work
/gemini:mode:auto "design microservices architecture for user management"
# → Auto-selects planning template

# 3. Manual template override when needed
/gemini:mode:auto "authentication issues" --template plan.md

# 4. List available templates
/gemini:mode:auto --list-templates

Intelligent Documentation Management

# 1. Daily development - context-aware updates
/update-memory                    # Default: related mode - detects and updates affected modules
/update-memory-related            # Explicit: context-aware updates based on recent changes

# 2. After working in specific module
cd src/api && /update-memory related  # Updates API module and parent hierarchy
/update-memory-related            # Same as above, with intelligent change detection

# 3. Periodic full refresh
/update-memory full               # Complete project-wide documentation update
/update-memory-full               # Explicit: full project scan with depth-parallel execution

# 4. Post-refactoring documentation sync
git commit -m "Major refactoring"
/update-memory-related            # Intelligently updates all affected areas with git-aware detection

# 5. Project initialization or major architectural changes
/update-memory-full               # Complete baseline documentation creation

Update Mode Comparison

Mode Trigger Complexity Threshold Best Use Case
related (default) Git changes + recent files >15 modules Daily development, feature work
full Complete project scan >20 modules Initial setup, major refactoring

📊 Complexity-Based Strategies

Complexity Task Count Hierarchy Depth File Structure Command Strategy
Simple <5 tasks 1 level (impl-N) Minimal structure Skip brainstorming → Direct implementation
Medium 5-15 tasks 2 levels (impl-N.M) Enhanced + auto-generated TODO_LIST.md Optional brainstorming → Action plan → Progress tracking
Complex >15 tasks 3 levels (impl-N.M.P) Complete document suite Required brainstorming → Multi-agent orchestration → Deep context analysis

🚀 v1.0 Release Benefits

  • Smart Automation: Intelligent template selection reduces manual template discovery
  • Documentation: 500+ lines streamlined while maintaining functionality
  • Template System: Dynamic discovery and YAML-based metadata parsing
  • Cross-Platform: Unified path handling for Windows/Linux compatibility
  • Developer Experience: Simplified commands with powerful auto-selection

🔧 Technical Highlights

  • Intelligent Context Processing: Dynamic context construction with technology stack detection
  • Template-Driven Architecture: Highly customizable and extensible through templates
  • Quality Assurance Integration: Built-in code review and testing strategy phases
  • Intelligent Documentation System: 4-layer hierarchical CLAUDE.md system with:
    • Dual-mode Operations: related (git-aware change detection) and full (complete project scan)
    • Complexity-adaptive Execution: Auto-delegation to memory-gemini-bridge for complex projects (>15/20 modules)
    • Depth-parallel Processing: Bottom-up execution ensuring child context availability for parent updates
    • Git Integration: Smart change detection with fallback strategies and comprehensive status reporting
  • CLI-First Design: Powerful, orthogonal command-line interface for automation

🎨 Design Philosophy

  • Structure over Freeform: Guided workflows prevent chaos and oversights
  • Traceability & Auditing: Complete audit trail for all decisions and changes
  • Automation with Human Oversight: High automation with human confirmation at key decision points
  • Separation of Concerns: Clean architecture with distinct responsibilities
  • Extensibility: Easy to extend with new agents, commands, and templates

📚 Documentation

  • Workflow Guidelines: See workflows/ directory for detailed process documentation
  • Agent Definitions: Check agents/ for AI agent specifications
  • Template Library: Explore planning-templates/ and prompt-templates/
  • Integration Guides: Review Gemini CLI integration in workflows/gemini-*.md

🤝 Contributing

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/amazing-feature
  3. Commit your changes: git commit -m 'Add amazing feature'
  4. Push to the branch: git push origin feature/amazing-feature
  5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🔮 Future Roadmap

  • Enhanced multi-language support
  • Integration with additional AI models
  • Advanced project analytics and insights
  • Real-time collaboration features
  • Extended CI/CD pipeline integration

Claude Code Workflow (CCW) - Transforming software development through intelligent automation and structured workflows.

Description
JSON-driven multi-agent development framework with intelligent CLI orchestration (Gemini/Qwen/Codex), context-first architecture, and automated workflow execution
Readme MIT 33 MiB
Languages
TypeScript 38.2%
Python 24.2%
HTML 16.3%
JavaScript 15.2%
CSS 4.5%
Other 1.6%