Unified MCP Tool Graph

v1.0.0Coding Agentsstable

Instead of dumping 1000+ tools into a model’s prompt and expecting it to choose wisely, the Unified MCP Tool Graph equips your LLM with structure, clarity, and relevance. It fixes tool confusion, prevents infinite loops, and enables modular, intellig

a2a-mcpa2a-protocolagentic-aiaiautomation
Share:
27
Stars
0
Downloads
0
Weekly
0/5

What is Unified MCP Tool Graph?

Unified MCP Tool Graph is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to instead of dumping 1000+ tools into a model’s prompt and expecting it to choose wisely, the unified mcp tool graph equips your llm with structure, clarity, and relevance. it fixes tool confusion, prev...

Instead of dumping 1000+ tools into a model’s prompt and expecting it to choose wisely, the Unified MCP Tool Graph equips your LLM with structure, clarity, and relevance. It fixes tool confusion, prevents infinite loops, and enables modular, intellig

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Instead of dumping 1000+ tools into a model’s prompt and exp

Use Cases

Structure tool discovery
Prevent infinite loops
Modular intelligence
LicenseMIT
Languagepython
Versionv1.0.0
UpdatedApr 16, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx unified-mcp-tool-graph

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Unified MCP Tool Graph

The Unified MCP Tool Graph is a Python-based MCP gateway that aggregates tools from over 4,000 MCP servers into a single graph-based index and serves only the most relevant tools to an LLM for any given task. Instead of overwhelming a model with thousands of tool definitions, it dynamically retrieves the top 3-5 relevant tools along with their server configurations on-demand, using a Neo4j-backed graph for semantic search and an mcp-proxy HTTP bridge for routing. It is designed for multi-agent systems where tool overload leads to confusion, infinite loops, or poor decisions.

Prerequisites

  • Python 3.10 or later with the uv package manager installed
  • Docker and Docker Compose (for running the Neo4j graph database and gateway services)
  • Approximately 4 GB of RAM for running the gateway and Neo4j locally
  • An MCP client such as Claude Desktop or any A2A-compatible agent runtime
1

Clone the repository

Clone the Unified MCP Tool Graph repository and navigate into it.

git clone https://github.com/pratikjadhav2726/Unified-MCP-Tool-Graph.git
cd Unified-MCP-Tool-Graph
2

Install dependencies with uv

Install all Python dependencies using the uv package manager as specified in the project.

uv sync
3

Start the unified gateway

Run the gateway startup script. This launches the MCP Unified Gateway on port 8000 and the mcp-proxy server on port 9000. Five popular MCP servers start immediately; others are spun up on-demand.

uv run python start_unified_gateway.py
# Gateway: http://localhost:8000
# Proxy:   http://localhost:9000
4

Query the gateway for relevant tools

Send a natural language task description to the gateway's tool retrieval endpoint. It returns the top 3-5 most relevant tools and the MCP server configurations needed to call them.

# Example HTTP request to the gateway:
curl -X POST http://localhost:8000/tools/search \
  -H 'Content-Type: application/json' \
  -d '{"query": "Schedule a LinkedIn post about AI trends"}'  
5

Configure your MCP client

Point your MCP client at the unified gateway's HTTP endpoint. The gateway exposes a standard MCP interface so any compatible client can connect.

{
  "mcpServers": {
    "unified-mcp-tool-graph": {
      "url": "http://localhost:8000/mcp"
    }
  }
}
6

Access individual servers via the mcp-proxy

Individual MCP servers managed by the gateway are also accessible through the mcp-proxy HTTP bridge, which lets you call a specific server's tools directly if needed.

# SSE endpoint for a specific server:
http://localhost:9000/servers/<server-name>/sse

# HTTP endpoint:
http://localhost:9000/servers/<server-name>/

Unified MCP Tool Graph Examples

Client configuration

Claude Desktop configuration connecting to the Unified MCP Tool Graph gateway via HTTP transport.

{
  "mcpServers": {
    "unified-mcp-tool-graph": {
      "url": "http://localhost:8000/mcp"
    }
  }
}

Prompts to try

These prompts demonstrate how the gateway surfaces only the relevant tools for each task rather than flooding the model with all 11,000+ available tools.

- "Schedule a LinkedIn post about the latest AI trends"
- "Search GitHub for open issues tagged 'good first issue' in the React repository"
- "Send a Slack message to the #engineering channel with today's deployment summary"
- "Look up the weather in Tokyo and add it to a Notion page"
- "Find the top 5 tools available for web scraping tasks"

Troubleshooting Unified MCP Tool Graph

Gateway fails to start with a port conflict on 8000 or 9000

Check if another process is already using these ports with `lsof -i :8000` and `lsof -i :9000`. Kill the conflicting process or edit start_unified_gateway.py to change the port numbers before starting.

Neo4j database is not persisting tool graph data between restarts

Neo4j is optional but recommended for graph persistence. Ensure the Docker volume for Neo4j is correctly mounted. Without Neo4j, the tool graph is rebuilt in memory on each startup, which adds initialization time.

On-demand MCP servers fail to spin up when needed

The gateway automatically spins up servers on-demand, but each server has its own dependencies. Check the gateway logs for the specific server that failed to start and ensure its required packages (npm packages, Python packages) are installed in the runtime environment.

Frequently Asked Questions about Unified MCP Tool Graph

What is Unified MCP Tool Graph?

Unified MCP Tool Graph is a Model Context Protocol (MCP) server that instead of dumping 1000+ tools into a model’s prompt and expecting it to choose wisely, the unified mcp tool graph equips your llm with structure, clarity, and relevance. it fixes tool confusion, prevents infinite loops, and enables modular, intellig It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Unified MCP Tool Graph?

Follow the installation instructions on the Unified MCP Tool Graph GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Unified MCP Tool Graph?

Unified MCP Tool Graph works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Unified MCP Tool Graph free to use?

Yes, Unified MCP Tool Graph is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "unified-mcp-tool-graph": { "command": "npx", "args": ["-y", "unified-mcp-tool-graph"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Unified MCP Tool Graph?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides