hive/mcp-server

🐝 WHOOSH MCP Server

Model Context Protocol (MCP) server that exposes the WHOOSH Distributed AI Orchestration Platform to AI assistants like Claude.

Overview

This MCP server allows AI assistants to:

• 🤖 Orchestrate Agent Tasks - Assign development work across your distributed cluster
• 📊 Monitor Executions - Track task progress and results in real-time
• 🔄 Manage Workflows - Create and execute complex distributed pipelines
• 📈 Access Cluster Resources - Get status, metrics, and performance data

Quick Start

1. Install Dependencies

cd mcp-server
npm install

2. Build the Server

npm run build

3. Configure Claude Desktop

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

Production (Swarm Deployment):

{
  "mcpServers": {
    "whoosh": {
      "command": "node",
      "args": ["/path/to/whoosh/mcp-server/dist/index.js"],
      "env": {
        "WHOOSH_API_URL": "https://whoosh.home.deepblack.cloud",
        "WHOOSH_WS_URL": "wss://whoosh.home.deepblack.cloud"
      }
    }
  }
}

Development/Local Testing:

{
  "mcpServers": {
    "whoosh": {
      "command": "node",
      "args": ["/path/to/whoosh/mcp-server/dist/index.js"],
      "env": {
        "WHOOSH_API_URL": "http://localhost:8087",
        "WHOOSH_WS_URL": "ws://localhost:8087"
      }
    }
  }
}

4. Restart Claude Desktop

The WHOOSH MCP server will automatically connect to your running WHOOSH cluster.

Available Tools

Agent Management

• whoosh_get_agents - List all registered agents with status
• whoosh_register_agent - Register new agents in the cluster

Task Management

• whoosh_create_task - Create development tasks for specialized agents
• whoosh_get_task - Get details of specific tasks
• whoosh_get_tasks - List tasks with filtering options

Workflow Management

• whoosh_get_workflows - List available workflows
• whoosh_create_workflow - Create new distributed workflows
• whoosh_execute_workflow - Execute workflows with inputs

Monitoring

• whoosh_get_cluster_status - Get comprehensive cluster status
• whoosh_get_metrics - Retrieve Prometheus metrics
• whoosh_get_executions - View workflow execution history

Coordination

• whoosh_coordinate_development - Orchestrate complex multi-agent development projects

Available Resources

Real-time Cluster Data

• whoosh://cluster/status - Live cluster status and health
• whoosh://agents/list - Agent registry with capabilities
• whoosh://tasks/active - Currently running and pending tasks
• whoosh://tasks/completed - Recent task results and metrics

Workflow Data

• whoosh://workflows/available - All configured workflows
• whoosh://executions/recent - Recent workflow executions

Monitoring Data

• whoosh://metrics/prometheus - Raw Prometheus metrics
• whoosh://capabilities/overview - Cluster capabilities summary

Example Usage with Claude

Register an Agent

Please register a new agent in my WHOOSH cluster:
- ID: walnut-kernel-dev
- Endpoint: http://walnut.local:11434  
- Model: codellama:34b
- Specialization: kernel_dev

Create a Development Task

Create a high-priority kernel development task to optimize FlashAttention for RDNA3 GPUs. 
The task should focus on memory coalescing and include constraints for backward compatibility.

Coordinate Complex Development

Help me coordinate development of a new PyTorch operator that includes:
1. CUDA/HIP kernel implementation (high priority)
2. PyTorch integration layer (medium priority)  
3. Performance benchmarks (medium priority)
4. Documentation and examples (low priority)
5. Unit and integration tests (high priority)

Use parallel coordination where possible.

Monitor Cluster Status

What's the current status of my WHOOSH cluster? Show me agent utilization and recent task performance.

Configuration

The MCP server connects to the WHOOSH backend using domain endpoints by default. You can customize this by setting environment variables:

Production (Default):

• WHOOSH_API_URL - https://whoosh.home.deepblack.cloud
• WHOOSH_WS_URL - wss://whoosh.home.deepblack.cloud

Development/Local Testing:

• WHOOSH_API_URL - http://localhost:8087
• WHOOSH_WS_URL - ws://localhost:8087

Additional Options:

• WHOOSH_TIMEOUT - Request timeout in milliseconds (default: 30000)

Copy .env.example to .env and modify as needed for your deployment.

Development

Watch Mode

npm run watch

Direct Run

npm run dev

Integration with WHOOSH

This MCP server connects to your running WHOOSH platform and provides a standardized interface for AI assistants to:

1. Understand your cluster capabilities and current state
2. Plan complex development tasks across multiple agents
3. Execute coordinated workflows with real-time monitoring
4. Optimize task distribution based on agent specializations

The server automatically handles task queuing, agent assignment, and result aggregation - allowing AI assistants to focus on high-level orchestration and decision-making.

Security Notes

• The MCP server connects to your local WHOOSH cluster
• No external network access required
• All communication stays within your development environment
• Agent endpoints should be on trusted networks only

---

🐝 Ready to let Claude orchestrate your distributed AI development cluster!