MCP Orchestrator

MCP Orchestrator

A unified interface that intelligently routes requests to appropriate MCP servers, solving the problem of managing multiple tools by providing a single entry point with smart routing capabilities.

Category
Visit Server

README

MCP Orchestrator 🎯

The ONE MCP to rule them all!

MCP Orchestrator solves the "too many tools" problem by providing a single, intelligent interface that helps Claude find the right tool among 100+ options. Instead of overwhelming Claude with all those tools, it provides smart routing guidance.

The Problem It Solves

When you have many MCP servers installed:

  • Claude gets confused with too many similar tools
  • It's hard to remember which tool does what
  • Authentication is scattered across configs
  • No central place to discover capabilities

The Solution

MCP Orchestrator acts as an intelligent tool finder:

  • Single Entry Point: Claude only sees 4 orchestrator tools
  • Smart Discovery: Natural language requests find the right tool
  • Tool Guidance: Returns which actual tool to use
  • Central Registry: All MCP capabilities in one place

How It Works

graph LR
    Claude[Claude] --> |"I need to generate an image"| Orchestrator[MCP Orchestrator]
    Orchestrator --> |"Use generate_image from comfyui"| Claude
    Claude --> |Calls actual tool| ComfyUI[ComfyUI MCP]

Important: The orchestrator doesn't execute tools itself - it tells Claude which tool to use!

Quick Start

  1. Install MCP Orchestrator:
git clone https://github.com/SamuraiBuddha/mcp-orchestrator.git
cd mcp-orchestrator
pip install -e .
  1. Start LM Studio with Granite embeddings model loaded

  2. Configure Claude Desktop to include the orchestrator alongside your other MCPs:

{
  "mcpServers": {
    "orchestrator": {
      "command": "python",
      "args": ["-m", "mcp_orchestrator"]
    },
    "comfyui": { ... },
    "github": { ... },
    "memory": { ... }
    // Keep all your existing MCPs!
  }
}
  1. Restart Claude Desktop

Now Claude has both the orchestrator AND all your regular tools available!

Core Tools

1. find_tool(query)

Discover which MCP and tool to use for a task:

# Ask: "I need to create a logo"
# Returns: 
# **comfyui** → generate_image
#   Confidence: 0.95
#   Description: Generate images using AI

2. execute(request, params)

Get routing information for any request:

# Ask: "generate a robot logo"
# Returns:
# To execute this request, use:
# **Tool**: generate_image
# **From MCP**: comfyui
# **Confidence**: 0.92
# 
# Use the actual `generate_image` tool from the MCP tools list to execute this.

3. list_capabilities(category)

See what's available:

# Ask: list_capabilities("image")
# Returns all image-related tools across all MCPs

4. explain_tool(mcp_name, tool_name)

Get detailed help:

# Ask: explain_tool("github", "create_repository")
# Returns parameters, examples, and best practices

What Makes It Special

🌍 Multilingual Support

Using IBM's Granite embeddings, the orchestrator understands requests in any language:

  • "Create a logo" 🇺🇸
  • "Créer un logo" 🇫🇷
  • "ロゴを作成" 🇯🇵
  • "创建标志" 🇨🇳

All route to the same tool!

🎯 Intelligent Routing

  • Uses semantic embeddings to understand intent
  • Handles paraphrases and variations
  • Provides confidence scores
  • Falls back gracefully when unsure

📈 Scales Infinitely

Add new MCPs to the registry and they're immediately discoverable. No need to retrain or reconfigure.

Architecture Benefits

  1. Reduced Cognitive Load: Claude focuses on 4 discovery tools instead of 100+ execution tools
  2. Better Tool Discovery: Find tools by describing what you need
  3. Central Registry: One place to document all capabilities
  4. Usage Analytics: See which tools provide the most value
  5. Easier Debugging: Know exactly which tool will be used

Real-World Example

Without Orchestrator:

  • Claude sees 100+ tools with confusing names
  • User: "Which one generates images???"
  • Claude: Tries to guess from tool names

With Orchestrator:

  • User: "Generate a Crisis Corps logo"
  • Orchestrator: "Use generate_image from comfyui MCP"
  • Claude: Calls the right tool directly

Configuration

The orchestrator is configured via config/registry.json. Each MCP entry includes:

  • Description and capabilities
  • Keywords for better matching
  • Tool definitions with examples

Future Enhancements

  • [ ] Web UI for managing MCP registry
  • [ ] Automatic MCP discovery
  • [ ] Natural language MCP creation
  • [ ] Integration with CORTEX for AI orchestration
  • [ ] Blockchain-based audit trail
  • [ ] Multi-agent coordination

The Vision

Imagine a world where:

  • Adding a new AI capability is as simple as installing an MCP
  • Claude automatically knows how to find and use it
  • No more tool overload or confusion
  • Perfect tool discovery every time

That's what MCP Orchestrator enables.

Contributing

This orchestrator is designed to grow with your MCP ecosystem. Add new MCPs to the registry and they're immediately discoverable through the unified interface.

License

MIT License - Use freely in your AI infrastructure!


Built with ❤️ for the AI community by someone who got tired of scrolling through 100+ tools

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured