Multi-Agent Squad 🤖

August 3, 2025 · View on GitHub

An intelligent orchestration system for Claude Code that adapts to your project structure and manages development through specialized AI agents.

🌟 Overview

Multi-Agent Squad transforms Claude Code into a complete software development team with enterprise-grade integrations. Simply start Claude in this directory, and it will:

  1. Ask you about your project through natural conversation
  2. Create the perfect structure based on your needs
  3. Set up specialized agents for your project type
  4. Configure integrations with 30+ popular tools
  5. Enable MCP servers for enhanced AI capabilities
  6. Orchestrate development through the entire lifecycle
  7. Automate workflows based on your specific needs

🎯 Complete System Architecture

┌──────────────────────────────────────────────────────┐
│                     MULTI-AGENT SQUAD                │
│                                                      │
│  ┌────────────────────┐    ┌─────────────────────┐   │
│  │   AI AGENTS        │    │   INTEGRATIONS      │   │
│  │                    │    │                     │   │
│  │ • Orchestrator     │    │ • Slack/Teams       │   │
│  │ • Product Mgr      │    │ • Jira/Linear       │   │
│  │ • Architect        │    │ • GitHub/GitLab     │   │
│  │ • Engineers        │    │ • CI/CD Tools       │   │
│  │ • QA/DevOps        │    │ • Monitoring        │   │
│  └────────────────────┘    └─────────────────────┘   │
│                                                      │
│  ┌────────────────────┐    ┌─────────────────────┐   │
│  │   MCP SERVERS      │    │   AUTOMATION        │   │
│  │                    │    │                     │   │
│  │ • Database         │    │ • Claude Hooks      │   │
│  │ • GitHub API       │    │ • Git Hooks         │   │
│  │ • Memory           │    │ • Sprint Mgmt       │   │
│  │ • Analytics        │    │ • PR Reviews        │   │
│  │ • Docker/K8s       │    │ • Quality Gates     │   │
│  └────────────────────┘    └─────────────────────┘   │
│                                                      │
└──────────────────────────────────────────────────────┘

🚀 Quick Start

# 1. Clone this repository
git clone https://github.com/bijutharakan/multi-agent-squad.git
cd multi-agent-squad

# 2. Start Claude Code
claude

# 3. Start the orchestration with ONE of these commands:
/project          # Recommended: Start project orchestration
"Start project"   # Natural language also works
"Help me set up"  # Any variation works

The /project command triggers the complete orchestration setup!

🌟 What You Get

After setup, you'll have:

  • 🤖 Specialized AI agents tailored to your project
  • 🔌 Configured integrations with your existing tools
  • 🚀 MCP servers for enhanced capabilities
  • 🦝 Automated workflows based on your needs
  • 📊 Sprint management with review cycles
  • 🔔 Smart notifications where you want them
  • 🔐 Security checks and quality gates

🎯 How It Works

First Time Setup

When you start Claude with /project, it will guide you through:

  1. "What are you building?" - Describe your project
  2. "Is this new or existing?" - It adapts accordingly
  3. "Monorepo or multi-repo?" - Your choice
  4. "How do you track tasks?" - GitHub, Jira, Linear, or manual
  5. "What would you like to automate?" - Custom hooks for your needs
  6. "Which integrations?" - Slack, email, Teams, and more
  7. "Enable MCP servers?" - Enhanced AI capabilities

Claude then creates the perfect structure, integrations, and automations for YOUR project.

Flexible Structure Examples

For a Multi-Repo Web App

your-project/
├── CLAUDE.md          # Orchestration instructions
├── PROJECT.md         # Your project details
├── projects/          # Your repos (you manage these)
│   ├── frontend/      # Your React app
│   └── backend/       # Your API
├── docs/              # All documentation
└── .claude/agents/    # Your AI team

For a Monorepo

your-project/
├── CLAUDE.md          # Orchestration instructions
├── PROJECT.md         # Your project details
├── src/               # All your code
│   ├── frontend/
│   ├── backend/
│   └── shared/
├── docs/              # Documentation
└── .claude/agents/    # Your AI team

For a Documentation Project

your-project/
├── CLAUDE.md          # Orchestration instructions
├── PROJECT.md         # Project details
├── docs/              # All documentation
│   ├── architecture/
│   ├── guides/
│   └── api/
└── .claude/agents/    # Documentation-focused agents

🎭 Your AI Development Team

Based on your project, Claude will suggest relevant agents:

For Full-Stack Projects

  • 🎯 Prime Orchestrator - Manages the entire workflow
  • 📊 Product Manager - Requirements and user stories
  • 🏗️ Solution Architect - System design
  • 💻 Backend Engineer - API development
  • 🎨 Frontend Engineer - User interface
  • ✅ QA Engineer - Testing
  • 🔧 DevOps Engineer - Deployment

For API Projects

  • 🏗️ API Architect - API design
  • 💻 Backend Engineer - Implementation
  • 📖 API Documentation Specialist - Docs
  • ✅ API Test Engineer - Testing

For Documentation Projects

  • 📝 Technical Writer - Documentation
  • 🏗️ Information Architect - Structure
  • 👁️ Documentation Reviewer - Quality

🛠️ Key Features

🤖 Enterprise Agile Workflow

  • Sprint Management - Automated ceremonies and tracking
  • PR Review Cycles - Enforced code review with comment resolution
  • Quality Gates - Automated checks before phase transitions
  • Human Checkpoints - Critical decisions require approval
  • Daily Standups - Automated reminders and reports

Natural Language Control

Just talk to Claude normally:

  • "I need to add user authentication"
  • "Help me set up CI/CD"
  • "Create API documentation"
  • "Review our architecture"

Intelligent Automation (Claude Code Hooks)

Tell Claude what you want automated:

  • "I forget to run tests" → Auto-runs tests after file changes
  • "Remind me of daily standups" → 9 AM Slack/email reminders
  • "Check code quality" → Auto-lint and format on save
  • "Track my progress" → Automatic time tracking and reports
  • "Security checks" → Scan for secrets before commits
  • "Break reminders" → Health reminders every hour

Hooks are dynamically generated based on YOUR specific needs using:

python scripts/generate-hooks.py

🔌 Extensive Integrations (30+ Tools)

Project Management

  • GitHub - Issues, PRs, Projects, Actions
  • Jira - Sprints, epics, stories, burndown charts
  • Linear - Modern issue tracking with cycles
  • Azure DevOps - Boards, repos, pipelines
  • ClickUp, Monday.com, Asana - Visual project management

Communication

  • Slack - Real-time notifications, daily summaries, thread management
  • Microsoft Teams - Adaptive cards, channel updates
  • Discord - Webhooks, embeds for communities
  • Email - HTML/plain text, daily digests, critical alerts

Development & DevOps

  • CI/CD - GitHub Actions, Jenkins, GitLab CI, CircleCI
  • Monitoring - Sentry, Datadog, Prometheus, New Relic
  • Testing - BrowserStack, SonarQube, Cypress Dashboard
  • Documentation - Confluence, Notion, GitHub Wiki
  • Time Tracking - Toggl, Harvest, Clockify

🚀 MCP (Model Context Protocol) Server Support

Extend Claude's capabilities with 14+ pre-configured MCP servers:

Data & Storage

  • PostgreSQL Explorer - Natural language database queries
  • Memory Server - Persistent context across Claude sessions
  • Enhanced Filesystem - Advanced file operations with permissions

Development Tools

  • GitHub Integration - Deep PR and issue management
  • Test Runner - Execute and monitor test suites
  • Docker Management - Container and image control
  • Kubernetes - Deployment and pod management

Communication & Docs

  • Slack Server - Read channels, send messages
  • Linear Server - Full issue tracking integration
  • Notion Server - Access pages and databases
  • Confluence Server - Documentation management

Analytics & Monitoring

  • Project Analytics - Custom metrics and insights
  • Monitoring Integration - Datadog, Prometheus access

Setup with: python scripts/mcp-server-setup.py

Smart Development Workflow

  1. Requirements → PM agent creates specs
  2. Design → Architect creates system design
  3. Implementation → Engineers build it
  4. Testing → QA ensures quality
  5. Deployment → DevOps handles release

Git Worktree Support (Optional)

For multi-repo projects with parallel development:

./scripts/worktree-manager.sh create-feature auth frontend backend

📋 No Complex Setup Required

Unlike other tools, Multi-Agent Squad:

  • ✅ No installation process - Just clone and start
  • ✅ No configuration files - Claude asks what you need
  • ✅ No forced structure - Adapts to your preferences
  • ✅ No learning curve - Plain English control

🔧 System Requirements

Required

  • Git - For version control
  • Claude Code - The AI interface

Optional

  • GitHub CLI (gh) - For GitHub integration
  • Python 3.8+ - For automation scripts
  • Jira API Token - For Jira integration

Claude will check what you have and work with it!

🔧 Available Scripts & Tools

Core Orchestration

  • /project - Main entry point to start orchestration
  • worktree-manager.sh - Git worktree management for multi-repo
  • discover-project.py - Analyze existing codebases

Integration Scripts

  • integration-setup.py - Universal integration manager
  • mcp-server-setup.py - MCP server configuration
  • agile-tools-setup.py - Agile tool integrations
  • slack-integration.py - Slack webhook setup
  • email-integration.py - Email notification setup
  • github-integration.py - GitHub API automation

Workflow Automation

  • sprint-management.sh - Sprint ceremonies and tracking
  • pr-review-cycle.sh - Automated PR review enforcement
  • generate-hooks.py - Dynamic hook generation
  • setup-git-hooks.sh - Git hook configuration

🎯 Example Workflows

Starting a New SaaS Product

You: "/project"
Claude: "Hello! I'm your Multi-Agent Squad Orchestrator. What are you building?"
You: "A SaaS product for team collaboration"
Claude: "Great! Is this a new project or existing code?"
You: "Brand new"
Claude: "How would you like to organize your code?"
[Shows options: monorepo, multi-repo, etc.]
You: "Multi-repo - separate frontend and backend"
Claude: "What would you like me to automate for you?"
You: "I often forget to run tests"
Claude: [Creates structure, agents, test automation hooks]

Setting Up Integrations

You: "Set up Slack notifications for builds"
Claude: "I'll help you set up Slack integration. Running the setup..."
[Interactive setup with permissions]
Claude: "Would you like notifications for:
  a) Build status
  b) Test results  
  c) Daily standup reminders
  d) PR reviews"
You: "All of them"
Claude: [Configures Slack with all requested notifications]

Using MCP Servers

You: "Show me all users in our database"
Claude: [Uses PostgreSQL MCP server]
"Here are the users in your database:
- alice@example.com (Admin)
- bob@example.com (User)
- charlie@example.com (User)"

You: "What's our test coverage?"
Claude: [Uses Analytics MCP server]
"Current test coverage: 78%
- Frontend: 82%
- Backend: 74%
Trend: +3% this week"

🎆 What Makes This Different?

🧠 True AI Orchestration

Unlike simple templates or scripts, Multi-Agent Squad uses Claude's intelligence to:

  • Understand your project contextually, not through rigid configs
  • Adapt to your workflow instead of forcing you into one
  • Learn your preferences and adjust automation accordingly
  • Make intelligent decisions while keeping you in control

🔗 Native Claude Code Integration

  • Built for Claude Code - Not a generic framework
  • Uses sub-agents - Real AI delegation, not just prompts
  • MCP Protocol - Direct tool access, not API wrappers
  • Claude Hooks - Deep integration with your workflow

🎯 Zero Configuration Philosophy

  • No YAML files - Just conversation
  • No setup wizards - Claude asks what it needs
  • No dependencies - Works with what you have
  • No lock-in - Your project, your way

🤝 Integration Setup

Multi-Agent Squad supports 30+ integrations across all aspects of agile development:

Quick Setup Commands

# Interactive integration setup (recommended)
python scripts/integration-setup.py

# Set up MCP servers for enhanced capabilities
python scripts/mcp-server-setup.py

# Configure specific agile tools
python scripts/agile-tools-setup.py

# Individual integrations
python scripts/slack-integration.py      # Slack notifications
python scripts/email-integration.py      # Email alerts
python scripts/github-integration.py     # GitHub automation

Integration Categories

  • 📋 Project Management - GitHub, Jira, Linear, Azure DevOps, ClickUp
  • 💬 Communication - Slack, Teams, Discord, Email
  • 📄 Documentation - Confluence, Notion, GitHub Wiki
  • 🚀 CI/CD - GitHub Actions, Jenkins, GitLab CI
  • 📊 Monitoring - Sentry, Datadog, Prometheus
  • 🧪 Testing - BrowserStack, SonarQube, Cypress
  • ⏱️ Time Tracking - Toggl, Harvest, Clockify

See docs/INTEGRATIONS.md for complete setup instructions.

📚 Advanced Features

🔐 Security & Permissions

  • Credential Management - Secure storage in .env.* files
  • Permission Scoping - Minimal required access
  • Secret Detection - Pre-commit hooks for security
  • MCP Server Security - Warnings for internet-accessing servers

Custom Agent Creation

You: "I need an agent who understands blockchain"
Claude: [Creates specialized blockchain agent]

MCP Server Capabilities

You: "Show me all users in the database"
Claude: [Uses PostgreSQL MCP server to query directly]

You: "What's our test coverage trend?"
Claude: [Uses Analytics MCP server to show metrics]

Multi-Environment Support

  • Development
  • Staging
  • Production
  • All managed through conversation

Automated Workflows

  • Code review orchestration
  • Deployment pipelines
  • Documentation updates
  • Sprint ceremonies
  • All through plain English

Critical Decision Gates

Claude always asks permission for:

  • Production deployments
  • Database migrations
  • Security changes
  • Main branch merges
  • Resource deletion

🛡️ Philosophy

Multi-Agent Squad believes in:

  • Flexibility - Your project, your way
  • Simplicity - No complex commands or configs
  • Intelligence - Agents with real expertise
  • Adaptability - Works with what you have
  • Transparency - Clear, documented decisions

🚦 Getting Help

Ask Claude

  • "How do I add a new feature?"
  • "What agents do I have?"
  • "Show me the project status"
  • "Help me set up Slack notifications"
  • "Configure MCP servers for my database"
  • "Create a custom automation hook"

Check Documentation

Quick Commands

# List available integrations
python scripts/integration-setup.py

# Check MCP server status
python scripts/mcp-server-setup.py check

# View configured tools
python scripts/agile-tools-setup.py list

📄 License

MIT License - see LICENSE file

🙏 Contributing

We welcome contributions! The best way to contribute is to use the system and share your experience.

Ready to start? Just cd into this directory and start Claude Code. No installation, no configuration, just natural conversation to build amazing software!

Multi-Agent Squad: Where AI meets human creativity to build exceptional software.