🎨 Dream-Weaver MCP

Autonomous Multi-Agent Project Builder - Transform ideas into reality with zero supervision

✨ What is Dream-Weaver?

Dream-Weaver is a revolutionary Model Context Protocol (MCP) server that transforms simple project ideas into complete, production-ready applications through autonomous multi-agent orchestration.

Core Philosophy:

• 🎯 Simple: Single request → Complete project
• 🤖 Autonomous: Zero supervision after approval
• ✅ Precise: Checkpoint-based validation prevents failures
• 📈 Scalable: From simple apps to smart city infrastructure
• 🔌 Flexible: Model-agnostic with dynamic routing

🚀 Quick Start

Installation

# Via NPX (recommended)
npx dream-weaver-mcp

# Or install globally
npm install -g dream-weaver-mcp

# Or via Python
pip install dream-weaver-mcp

Configuration

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "dream-weaver": {
      "command": "npx",
      "args": ["-y", "dream-weaver-mcp"],
      "env": {
        "OPENROUTER_API_KEY": "your-key-here",
        "MODEL_ROUTER": "litellm",
        "DEFAULT_MODEL": "anthropic/claude-3.5-sonnet"
      }
    }
  }
}

Environment Variables

Create a .env file:

MODEL_ROUTER=litellm
OPENROUTER_API_KEY=your-openrouter-key
DEFAULT_MODEL=anthropic/claude-3.5-sonnet
COST_OPTIMIZATION=true
LOG_LEVEL=info

📚 Usage

1. Initialize Your Dream

// In Claude or any MCP client
await use_mcp_tool("dream_init", {
  project_idea: "Build a real-time collaborative whiteboard app"
});

Response:

🎨 Dream-Weaver Initialized

Conversation ID: abc-123-xyz

🎯 Let's refine your vision!

I understand you want to build: "Real-time collaborative whiteboard app"

To create the perfect blueprint, I need to understand:
1. Target Users: Who will use this?
2. Key Features: What are the 3-5 most important features?
3. Scale: Expected user count?
...

2. Respond to Clarifications

await use_mcp_tool("dream_respond", {
  conversation_id: "abc-123-xyz",
  response: `
    - Target: Design teams and educators
    - Key Features: Real-time drawing, video chat, templates, export to PDF
    - Scale: 100-1000 concurrent users
    - Timeline: 4-6 weeks
    - Tech: Modern web stack, prefer React
  `
});

3. Approve & Launch Autonomous Build

await use_mcp_tool("dream_approve", {
  conversation_id: "abc-123-xyz",
  approved: true
});

Response:

🚀 Autonomous Build Started!

Project ID: project-456

Teams Assembled:
- Frontend Team: 3 agents + 1 PM
- Backend Team: 3 agents + 1 PM
- DevOps Team: 2 agents + 1 PM
- QA Team: 2 agents + 1 PM

Phases:
1. Foundation (1 week)
2. Development (2 weeks)
3. Integration (1 week)
4. Deployment (3 days)

Building will proceed autonomously with checkpoint validation!

4. Monitor Progress

await use_mcp_tool("dream_status", {
  project_id: "project-456"
});

🏗️ Architecture

System Layers

┌─────────────────────────────────────────┐
│     User Interface Layer (MCP)          │
│  Simple JSON Schema | NPM Installation  │
└─────────────────────────────────────────┘
              ↓
┌─────────────────────────────────────────┐
│      Orchestration Layer                │
│  Conversation | Planning | Spawning     │
└─────────────────────────────────────────┘
              ↓
┌─────────────────────────────────────────┐
│      Planning Layer                     │
│  Specs | Architecture | Blueprints      │
└─────────────────────────────────────────┘
              ↓
┌─────────────────────────────────────────┐
│      Execution Layer                    │
│  PMs (1-40+) | Teams | Specialists      │
└─────────────────────────────────────────┘
              ↓
┌─────────────────────────────────────────┐
│      Validation Layer                   │
│  Checkpoints | Tests | Quality          │
└─────────────────────────────────────────┘

Workflow

1. Conversational Initialization → Understand requirements
2. Mastermind Planning → Enhance vision, create blueprint
3. Team Assembly → Spawn PMs and specialized agents
4. Parallel Execution → Teams work autonomously
5. Checkpoint Validation → Zero-failure guarantee
6. Final Delivery → Production-ready project

🛠️ Available Tools

| Tool | Description | |------|-------------| | dream_init | Start conversational project initialization | | dream_respond | Continue conversation with clarifications | | dream_approve | Approve blueprint and start autonomous build | | dream_status | Get real-time project status | | dream_list | List all projects and their status | | dream_configure | Configure model routing and preferences |

📊 Scalability

Small Project (TODO App)

• 1 PM + 5 Agents
• Timeline: 2-4 hours
• Teams: Frontend, Backend, QA

Medium Project (E-commerce)

• 5 PMs + 19 Agents
• Timeline: 1-2 days
• Teams: Frontend, Backend, Database, DevOps, QA, Security

Large Project (Smart City)

• 40 PMs + 95 Agents
• Timeline: 1-2 weeks
• Teams: IoT, Data Processing, Analytics, Infrastructure, Security, Integration, QA, DevOps

🔧 Technical Stack

• MCP Implementation: TypeScript SDK, Python SDK, FastMCP
• Agent Frameworks: Swarms, CrewAI, Custom orchestration
• Model Routing: LiteLLM, OpenRouter, Ollama
• Communication: Redis, WebSocket, gRPC
• Storage: PostgreSQL, Redis, S3
• Monitoring: Prometheus, Grafana

🎯 Key Features

✨ Conversational initialization - No complex forms 🧠 Mastermind planning - Enhances your vision ⚡ Dynamic team spawning - Based on complexity 🔄 Checkpoint validation - Prevents failures 🎯 Zero mock data - Production-ready output 🌐 Seamless scaling - Apps to infrastructure 🔌 Model-agnostic - Flexible routing 📦 Simple setup - Easy installation 🚀 Autonomous execution - After approval 🎨 Exceeds expectations - Every time

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE for details.

🌟 Credits

Built with inspiration from:

• Swarms - Hierarchical agent orchestration
• CrewAI - Role-playing autonomous agents
• LiteLLM - Universal LLM API
• MCP - Model Context Protocol

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Discord: Join our community

Made with ❤️ by the Dream-Weaver Team