MCP Server Proto-OKN

A Model Context Protocol (MCP) server that provides tools for querying SPARQL endpoints, with specialized support for the NSF-funded Proto-OKN Project (Prototype Open Knowledge Network) knowledge graphs hosted on the FRINK platform.

Features

• FRINK Integration: Automatic detection and documentation linking for FRINK-hosted knowledge graphs
• Proto-OKN Support: Optimized for querying knowledge graphs in the Proto-OKN ecosystem including:
  • SPOKE (Scalable Precision Medicine Open Knowledge Engine)
  • BioBricks ICE (Chemical safety and cheminformatics)
  • DREAM-KG (Addressing homelessness with explainable AI)
  • SAWGraph (Safe Agricultural Products and Water monitoring)
  • And many other Proto-OKN knowledge graphs
• Flexible Configuration: Support for both FRINK and custom SPARQL endpoints
• Automatic Documentation: Registry links for supported knowledge graphs

Overview

Installation

Prerequisites

1. Install VS Code Insiders (required for MCP support)

   Download and install VS Code Insiders from https://code.visualstudio.com/insiders/

   VS Code Insiders is needed because it includes the latest MCP (Model Context Protocol) features.

2. Install GitHub Copilot extension (required for MCP integration)

   • Open VS Code Insiders
   • Sign in with your GitHub account that has Copilot access
   • Note: You need an active GitHub Copilot subscription to use MCP features

   MCP servers integrate with VS Code through the Copilot Chat interface.

3. Install uv (Python package manager)

   # macOS/Linux
   curl -LsSf https://astral.sh/uv/install.sh | sh
   
   # Windows
   powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
   
   # Or via pip
   pip install uv

Setup Instructions

1. Clone and setup the project

   git clone https://github.com/sbl-sdsc/mcp-proto-okn.git
   cd mcp-proto-okn
   uv sync

2. Configure the MCP servers

   This project includes a pre-configured .vscode/mcp.json file with multiple Proto-OKN knowledge graph endpoints. You can add third party endpoints with a custom description as shown for Wikidata in the example.

   Edit .vscode/mcp.json to update the server configurations. Here is a snippet of the mcp.json file:

   {
     "servers": {
       "mcp-spoke-sparql": {
         "command": "uv",
         "args": ["run", "python", "-m", "mcp_server_protookn.server", "--endpoint", "https://frink.apps.renci.org/spoke/sparql"]
       },
      "mcp-wikidata-sparql": {
         "command": "uv",
         "args": ["run", "python", "-m", "mcp_server_protookn.server", "--endpoint", "https://query.wikidata.org/sparql", "--description", "Access to Wikidata's knowledge graph"]
      }
     }
   }

   The existing file contains configurations for a subset of Proto-OKN knowledge graphs. You can enable/disable specific servers by adding or removing them from the configuration.

3. Start using the MCP server

   • Open VS Code Insiders
   • File -> Open Folder: mcp-proto-okn
   • Open a new chat window
   • Select Agent mode
   • Select the Claude Sonnet 4 model (all other models including GPT-5 perform poorly)
   • The MCP servers should automatically connect and provide access to the knowledge graphs

Quick Start: Query a Knowledge Graph

Once everything is set up, you can start querying knowledge graphs through the VS Code chat interface:

Example prompts to try:

1. Get a description of a knowledge graph:

   Provide a concise overview of the SPOKE knowledge graph, including its main purpose, data sources, and key features.
   

2. Query that combines multiple entity types:

   Antibiotic contamination can contribute to antimicrobial resistance. Find locations with antibiotic contamination.
   

3. Query across multiple KGs:

   What type of data is available for perfluorooctanoic acid in SPOKE, BioBricks, and SAWGraph?
   

The chat interface will use the MCP server to execute SPARQL queries against the configured endpoints and return structured results.

Usage

Command Line Parameters

The MCP server accepts the following command line arguments:

Required:

• --endpoint: SPARQL endpoint URL (e.g., https://frink.apps.renci.org/spoke/sparql)

Optional:

• --description: Custom description for the SPARQL endpoint (automatically generated for FRINK endpoints)

Tools

query

Execute a SPARQL query against the configured endpoint.

Parameters:

• query_string: A valid SPARQL query string

Returns:

• The query results in JSON format

get_description

Get a description and other metadata about the endpoint, including the PI, funding information, and more.

Parameters:

None

Returns:

• A description of the SPARQL endpoint extracted from the ProtOKN registry or a custom description provided in mcp.json configuration file.

Links

• Proto-OKN Project
• FRINK Platform
• Knowledge Graph Registry
• MCP Protocol
• Original MCP Server SPARQL

Citation

PW Rose, CA Nelson, Y Shi, SE Baranzini, MCP Server Proto-OKN. Available online: https://github.com/sbl-sdsc/mcp-proto-okn (2025)

PW Rose, CA Nelson, SG Gebre, K Soman, KA Grigorev, LM Sanders, SV Costes, SE Baranzini, NASA SPOKE-GeneLab Knowledge Graph. Available online: https://github.com/BaranziniLab/spoke_genelab (2025)

CA Nelson, PW Rose, K Soman, LM Sanders, SG Gebre, SV Costes, SE Baranzini, Nasa Genelab-Knowledge Graph Fabric Enables Deep Biomedical Analysis of Multi-Omics Datasets, https://ntrs.nasa.gov/citations/20250000723 (2025)

L Sanders, S Costes, K Soman, P Rose, C Nelson, A Sawyer, S Gebre, S Baranzini, Biomedical Knowledge Graph Capability for Space Biology Knowledge Gain, 45th COSPAR Scientific Assembly. Held 13-21 July, 2024, https://ui.adsabs.harvard.edu/abs/2024cosp...45.2183S/abstract

Funding

NSF Award number 2333819, Proto-OKN Theme 1: Connecting Biomedical information on Earth and in Space via the SPOKE knowledge graph.