🇲🇰 mk-data-mcp

Bridging the Gap Between AI Agents and Macedonian Public Data

The Macedonian Public Data MCP Server is a standardized data wrapper for official Macedonian government institutional APIs. It allows AI models (like Claude, GPT, and Gemini) to move beyond their training data by providing real-time, authoritative access to public records through the Model Context Protocol (MCP).

🎯 Vision & Mission

The goal of this project is to provide a universal, standardized interface for public Macedonian data. By exposing these datasets as MCP Tools, any AI agent or client can easily query and consume them, empowering:

• Data Scientists: To access clean, structured datasets for analysis and visualization.
• Journalists & Students: To fact-check and analyze trends in Macedonia.
• Citizens: To easily access public information without navigating complex government portals.
• Public Administration: To support evidence-based decision-making.

🚀 Core Functionalities

1. Natural Language Querying: Users can ask questions like "What was the unemployment rate in the last 10 years?" and the AI will use the correct tool to fetch the answer.
2. Dataset Discovery: Search and browse datasets from multiple official sources.
3. Data Normalization: Varied source formats (CSV, XLS, multi-dimensional JSON) are transformed into a unified, flat JSON schema.
4. Source Attribution: Responses always include original source URLs and metadata, ensuring transparency.
5. Remote Transport: Optimised for serverless environments with HTTP/SSE transport.

📊 Primary Data Sources

The server integrates with established government repositories:

• MakStat (State Statistical Office): The authoritative source for population, demographics, and economic statistics.
• data.gov.mk (National Open Data Portal): A decentralized repository for government datasets across health, education, transport, and more.
• [Future Integrations]: Planned support for Open Finance, legislative data from Sobraine (Assembly), and municipal portals.

🏗️ System Architecture

Built on the Remote MCP Server paradigm, the project leverages modern web technologies for scalability and performance.

Technical Stack:

• Framework: Next.js 16 (App Router)
• Language: TypeScript (Strict mode)
• SDK: @modelcontextprotocol/sdk
• Transport: HTTP / Server-Sent Events (SSE)
• Styling: Tailwind CSS 4 & Shadcn/UI
• Deployment: Optimized for Vercel (Serverless)

Request Pipeline:

1. Tool Invocation: Client triggers a tool (e.g., get_makstat_demographics).
2. Upstream Request: Server fetches live data from official government APIs.
3. Normalization: Raw data is processed into a standardized JSON format.
4. Response: Clean results are returned to the AI client.

🛠️ Available Tools

🚦 Getting Started

Local Development

1. Clone the repository:

   git clone https://github.com/stefantrajanov/mk-data-mcp.git
   cd mk-data-mcp
   

2. Install dependencies:

   npm install
   # or
   bun install
   

3. Run the development server:

   npm run dev
   # or
   bun dev
   

4. Access the MCP endpoint: The server listens on http://localhost:3000/api/mcp.

Deployment

The project is designed to be deployed to Vercel with zero configuration. It uses the Next.js Edge Runtime for high-performance, stateless execution.

📡 Connecting to Your AI Client

To use this server in your preferred MCP client (like Claude Desktop or Cursor), point the client to the SSE endpoint:

 "mcpServers": {
    "mk-data-mcp": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mk-data-mcp.vercel.app/api/mcp/mcp"]
    }
  },

For local testing, use: http://localhost:3000/api/mcp/mcp

✨ Design Principles

The project's UI/UX (for future dashboards/visualizations) follows:

• Typography: Inter (Modern sans-serif) for high readability.
• Aesthetics: High-contrast slate text, subtle gradients, and reactive micro-animations.
• Interoperability: Full compatibility with the official MCP SDK and modern AI interfaces.

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.