Graph Tools

You can run Graph Tools as an MCP server to analyze, visualize, and extract insights from graphs using HTTP REST API or Claude Desktop. This MCP server combines a Ruby core for graph operations with a Node.js-based MCP interface, letting you generate adjacency matrices, perform DFS/BFS, compute centrality, and visualize results programmatically or in a GUI.

How to use

You connect your MCP client to the Graph Tools MCP server via two available methods: HTTP REST API or Claude Desktop integration. Use the HTTP endpoint for automated workflows and remote clients, or configure Claude Desktop to drive graph analysis through natural language requests. The server exposes endpoints for analyzing relationships, building adjacency matrices, calculating centrality, and performing comprehensive network analyses.

How to install

Prerequisites you need before starting:

• Ruby 2.7+ for core graph operations
• Node.js 16+ for MCP server functionality
• A modern web browser for visualizations

Step-by-step setup flow you can follow to run the MCP server locally:

# 1) Clone the project repository
git clone https://github.com/dromologue/Graph-Tools.git
cd Graph-Tools

# 2) Install Ruby dependencies for core graph operations
gem install

# 3) Install MCP server dependencies
cd mcp-graph-server
npm install

# 4) Return to the project root for web app setup
cd ..

# 5) Install web application dependencies
npm install

Additional configuration and run instructions

The MCP server supports two connection methods. You can run the server as an HTTP REST API or as a local STDIO-based server for Claude Desktop integration. The following configurations are shown in examples:

{
  "mcpServers": [
    {
      "type": "http",
      "name": "graph_http",
      "url": "http://localhost:3001/",
      "args": []
    },
    {
      "type": "stdio",
      "name": "graph_claude",
      "command": "node",
      "args": ["/path/to/Graph-Tools/mcp-graph-server/api-server.js"],
      "env": {
        "SERVER_MODE": "mcp"
      }
    }
  ]
}

Security and best practices

• Keep your MCP API endpoints secured behind appropriate authentication if exposed beyond a trusted network.
• Use environment variables to configure runtime modes (for example, SERVER_MODE) and avoid placing sensitive data in code or repository files.
• Ensure Node.js and Ruby versions are kept up to date to benefit from security patches and performance improvements.

Troubleshooting

• If the HTTP endpoint does not respond, verify that the MCP server is running and accessible at the configured URL. Check that port 3001 is not blocked by a firewall.
• If Claude Desktop cannot start the MCP workflow, ensure the Claude config points to the correct node script and supports the SERVER_MODE=mcp environment variable.
• Review installation steps to confirm all dependencies are installed before attempting to start servers.

Notes

• The MCP server provides an HTTP REST API and a Claude Desktop integration path. You can interoperate between programmatic clients and natural language-driven workflows.
• Core graph operations, visualization exports, and centrality analysis are available through the API, enabling end-to-end graph analysis pipelines.

API endpoints overview

The MCP server exposes endpoints for common graph analytics tasks, including analyzing relationships, creating adjacency matrices, calculating centrality, and performing comprehensive network analyses. Use these endpoints from your MCP client to drive data-driven graph insights.

Available tools

analyze_relationships

Extract relationships from structured data and produce visualizations of graph connections.

create_adjacency_matrix

Build adjacency matrices from provided relationship pairs for analysis and visualization.

calculate_centrality

Compute centrality measures such as degree, betweenness, closeness, and eigenvector for nodes in the graph.

analyze_network_structure

Perform comprehensive network analysis combining data extraction and centrality calculations.