ClaudeMod
Back to browse
Harnesses

Claude Code Workflow

catlog22's workflow harness for product teams β€” sprint-based task management, issue-to-code automation, and structured PR generation workflows.

catlog221,697 starsAdded 1 months ago

Version npm License Stars Issues

English | δΈ­ζ–‡


Typing SVG


Quick Start Guide CLI Tools Architecture


✨ Key Features

🎯 Skill-based Workflows

From lite-plan (lightweight) to brainstorm (multi-role analysis)

πŸ”„ Multi-CLI Orchestration

Gemini, Qwen, Codex, Claude - auto-select or manual

⚑ Team Architecture v2

Role-based agents with inner loop execution

πŸ”§ Queue Scheduler

Background queue execution service

πŸ“¦ Session Lifecycle

start/resume/complete/sync workflow sessions

πŸ–₯️ Terminal Dashboard

Multi-terminal grid with execution monitor

🎨 Orchestrator Editor

Template-based workflow visual editing

πŸ’¬ A2UI

Agent-to-User interactive interface

πŸ“– New? See Workflow Guide for complete workflow documentation.


πŸš€ Quick Start

Install CCW

npm install -g claude-code-workflow
ccw install -m Global

Codex Configuration (Required for .codex/skills/)

If you use Codex CLI with the .codex/skills/ workflow skills, add these required features to your ~/.codex/config.toml:

[features]
default_mode_request_user_input = true   # Enable request_user_input tool for interactive confirmations
multi_agent = true                       # Enable multi-agent coordination (spawn_agent, wait, etc.)
multi_agent_v2 = true                    # Enable v4 agent API (fork_context, task_name, send_message, assign_task, list_agents)
enable_fanout = true                     # Enable spawn_agents_on_csv for parallel wave execution

These features are required for workflow skills to function properly. Without them, interactive confirmation gates (request_user_input), subagent orchestration, and CSV-driven parallel execution will not work.

Workflow Skills vs Commands

CCW uses two types of invocations:

TypeFormatExamples
SkillsTrigger phrase (no slash)workflow-lite-plan, brainstorm, workflow-plan
CommandsSlash command/ccw, /workflow/session:start, /issue/new

Choose Your Workflow Skill

Skill TriggerUse Case
workflow-lite-planLightweight planning, single-module features (hands off to lite-execute)
workflow-multi-cli-planMulti-CLI collaborative analysis
workflow-planFull planning with session persistence
workflow-tdd-planTest-driven development
workflow-test-fixTest generation and fix cycles
brainstormMulti-role brainstorming analysis

Workflow Examples

# Skill triggers (no slash - just describe what you want)
workflow-lite-plan "Add JWT authentication"
workflow-plan "Implement payment gateway integration"
workflow-execute

# Brainstorming
brainstorm "Design real-time collaboration system"

# Slash commands for session management
/workflow/session:start
/workflow/session:resume
/workflow/session:complete

πŸ› οΈ CLI Tool Installation

CLIDescriptionOfficial Docs
GeminiGoogle AI analysisgoogle-gemini/gemini-cli
CodexOpenAI autonomous codingopenai/codex
OpenCodeOpen-source multi-modelopencode-ai/opencode
QwenAlibaba Qwen-CodeQwenLM/Qwen

🎭 Semantic CLI Invocation


Users can semantically specify CLI tools in prompts - the system automatically invokes the corresponding CLI.

Basic Invocation

User PromptSystem Action
"Use Gemini to analyze the auth module"Auto-invoke gemini CLI for analysis
"Let Codex review this code"Auto-invoke codex CLI for review
"Ask Qwen about performance optimization"Auto-invoke qwen CLI for consultation

Multi-CLI Orchestration

PatternUser Prompt Example
Collaborative"Use Gemini and Codex to collaboratively analyze security vulnerabilities"
Parallel"Have Gemini, Codex, and Qwen analyze the architecture in parallel"
Iterative"Use Gemini to diagnose, then Codex to fix, iterate until resolved"
Pipeline"Gemini designs the solution, Codex implements, Claude reviews"

πŸ” ACE Tool Configuration

ACE (Augment Context Engine) provides powerful semantic code search.


πŸ“š CodexLens Local Search

⚠️ In Development: CodexLens is under iterative optimization. Some features may be unstable.

Search ModeDescription
FTSFull-text search, based on SQLite FTS5
SemanticSemantic search, using local embedding models
HybridHybrid search, combining FTS + Semantic + Reranking

πŸ’» CCW CLI Commands

🌟 Recommended Commands

CommandDescriptionWhen to Use
/ccwAuto workflow orchestrator - analyzes intent, selects workflow, executesβœ… General tasks, auto workflow selection
/ccw-coordinatorSmart orchestrator - recommends command chains, allows manual adjustmentπŸ”§ Complex multi-step workflows

Quick Examples:

# /ccw - Auto workflow selection
/ccw "Add user authentication"
/ccw "Fix memory leak in WebSocket"
/ccw "Implement with TDD"

# /ccw-coordinator - Manual chain orchestration
/ccw-coordinator "Implement OAuth2 system"

Session Management Commands

/workflow:session:start     # Start new workflow session
/workflow:session:resume    # Resume paused session
/workflow:session:list      # List all sessions
/workflow:session:sync      # Sync session work
/workflow:session:complete  # Complete session

Issue Workflow Commands

/issue/new       # Create new issue
/issue/plan      # Plan issue resolution
/issue/queue     # Form execution queue
/issue/execute   # Execute issue queue

Other CLI Commands

ccw install           # Install workflow files
ccw view              # Open dashboard
ccw cli -p "..."      # Execute CLI tools (Gemini/Qwen/Codex)
ccw upgrade -a        # Upgrade all installations

πŸ—οΈ Architecture

β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚                     Workflow Skills                              β”‚
β”‚  πŸ“ workflow-lite-plan / workflow-multi-cli-plan (lightweight)  β”‚
β”‚  πŸ“Š workflow-plan / workflow-tdd-plan (session-based)           β”‚
β”‚  πŸ§ͺ workflow-test-fix / workflow-test-fix         β”‚
β”‚  🧠 brainstorm (multi-role analysis)                            β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                              ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚                   Team Architecture v2                          β”‚
β”‚  πŸ€– team-worker agents with role-spec based execution           β”‚
β”‚  πŸ”„ Inner loop framework for sequential task processing         β”‚
β”‚  πŸ“’ Message bus protocol with team coordination                 β”‚
β”‚  🧠 Wisdom accumulation (learnings/decisions/conventions)       β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                              ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚                   Queue Scheduler Service                       β”‚
β”‚  βš™οΈ Background execution service with API endpoints             β”‚
β”‚  πŸ“Š Queue management and unified CLI execution settings         β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                              ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚                   Terminal Dashboard & Orchestrator             β”‚
β”‚  πŸ–₯️ Multi-terminal grid with execution monitor                  β”‚
β”‚  🎨 Template-based workflow editor with slash commands          β”‚
β”‚  πŸ“‘ Real-time agent communication via A2UI                      β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

πŸ“¦ Project Structure

Claude-Code-Workflow/
β”œβ”€β”€ .claude/
β”‚   β”œβ”€β”€ agents/          # 22 specialized agents (team-worker, cli-discuss, etc.)
β”‚   β”œβ”€β”€ commands/        # Slash commands (5 categories)
β”‚   β”‚   β”œβ”€β”€ ccw.md       # Main orchestrator
β”‚   β”‚   β”œβ”€β”€ ccw-coordinator.md
β”‚   β”‚   β”œβ”€β”€ cli/         # CLI commands (cli-init, codex-review)
β”‚   β”‚   β”œβ”€β”€ issue/       # Issue management (plan, execute, queue)
β”‚   β”‚   β”œβ”€β”€ memory/      # Memory commands (prepare, style-skill-memory)
β”‚   β”‚   └── workflow/    # Workflow commands (session, ui-design, etc.)
β”‚   └── skills/          # 37 modular skills
β”‚       β”œβ”€β”€ workflow-lite-plan/
β”‚       β”œβ”€β”€ workflow-plan/
β”‚       β”œβ”€β”€ workflow-tdd-plan/
β”‚       β”œβ”€β”€ workflow-test-fix/
β”‚       β”œβ”€β”€ brainstorm/
β”‚       β”œβ”€β”€ team-*/      # Team coordination skills
β”‚       └── ...
β”œβ”€β”€ ccw/
β”‚   β”œβ”€β”€ src/             # TypeScript source code
β”‚   β”‚   β”œβ”€β”€ commands/    # CLI command implementations
β”‚   β”‚   β”œβ”€β”€ core/        # Core services (a2ui, auth, hooks, routes)
β”‚   β”‚   β”œβ”€β”€ mcp-server/  # MCP server implementation
β”‚   β”‚   └── tools/       # Tool implementations
β”‚   └── frontend/        # React frontend (Terminal Dashboard, Orchestrator)
β”œβ”€β”€ codex-lens/          # Local semantic code search engine
└── docs/                # Documentation

🎼 Team Cadence Control (Beat Model)

The v2 team architecture introduces an event-driven beat model for efficient orchestration:

Beat Cycle (single beat)
======================================================================
  Event                   Coordinator              Workers
----------------------------------------------------------------------
  callback/resume --> +- handleCallback -+
                      |  mark completed   |
                      |  check pipeline   |
                      +- handleSpawnNext -+
                      |  find ready tasks |
                      |  spawn workers ---+--> [team-worker A] Phase 1-5
                      |  (parallel OK)  --+--> [team-worker B] Phase 1-5
                      +- STOP (idle) -----+         |
                                                     |
  callback <-----------------------------------------+
  (next beat)              SendMessage + TaskUpdate(completed)
======================================================================

Key Benefits:

  • 🎯 Event-driven: Coordinator only wakes when needed (callback/resume)
  • ⚑ Fast-advance: Simple successors spawn directly without coordinator roundtrip
  • πŸ”„ Dynamic pipelines: Generated per-task from dependency graph
  • πŸ“Š Parallel execution: Independent tasks run concurrently

πŸ–₯️ Frontend Highlights

Terminal Dashboard

Multi-terminal grid layout with real-time execution monitoring.

Features:

  • πŸ–₯️ Multi-terminal grid with resizable panes
  • πŸ“Š Execution monitor with agent list
  • πŸ“ File sidebar for project navigation
  • 🎯 Session grouping by project tags
  • πŸŒ™ Fullscreen/immersive mode

Orchestrator Editor

Visual workflow template editor with drag-drop.

Features:

  • 🎨 React Flow-based visual editing
  • πŸ“¦ Template library with pre-built workflows
  • πŸ”§ Property panel for node configuration
  • ⚑ Slash command integration

🀝 Contributing

GitHub Issues

πŸ“„ License

MIT License - see LICENSE


πŸ”— Links

LINUX DO

More Harnesss

Harnesses

claude-simone

Structured project management harness for Claude Code β€” define milestones and sprints in markdown, then run execution loops that drive Claude through each task.

project-managementmilestonessprints+2
by Helmi
GitHub
Harnesses

Claude Squad

Multi-agent tmux orchestrator β€” spawn N isolated Claude Code sessions in parallel, each in its own git worktree, then merge the best results back to main.

multi-agentparalleltmux+3
by SMTG AI
GitHub
Harnesses

Claude Swarm

Distributed Claude agent swarms β€” agents communicate via MCP, share tool access, and collaborate on complex tasks with configurable topology and role specialization.

swarmdistributedmulti-agent+3
by parruda
GitHub

Command Palette

Search for a command to run...