kolada_mcp by dennisjernkrookratio - MCP Server

Kolada MCP Server# Kolada MCP Server

Model Context Protocol (MCP) server exposing Sweden's Kolada open data using the koladapy client. The server provides search and retrieval helpers suitable for Deep Research and other MCP-compatible clients.

FeaturesKolada MCP Server enables seamless integration between Large Language Models (LLMs) and Kolada, Sweden’s comprehensive municipal and regional statistical database. It provides structured access to thousands of Key Performance Indicators (KPIs), facilitating rich, data-driven analysis, comparisons, and explorations of public sector statistics.

Discover KPIs and municipalities via semantic and keyword search
Fetch detailed metric values from Kolada's REST API## Overview
Built on fastmcp for straightforward SSE deployment

Kolada MCP Server acts as intelligent middleware between LLM-based applications and the Kolada database, allowing easy querying and analyzing of data related to Swedish municipalities and regions. With semantic search capabilities and robust analysis tools, Kolada MCP significantly simplifies navigating and interpreting the vast array of KPIs in Kolada.

Quick start


pip install .

kolada-mcpAsk Kolada MCP Server complex questions requiring data analysis:

```- Where in Sweden should a family move to find affordable housing, good schools, and healthcare?

- Investigate the connection between unemployment and mental illness in Västernorrland.

The server listens on the port defined by the `PORT` environment variable (defaults to `8000`). Clients should connect to the `/sse` endpoint for Server-Sent Events.- Identify municipalities with the highest increase in preschool quality over the last five years.

- Create a dashboard visualizing municipalities with the best and worst public transportation.

## Features
- **Semantic Search**: Natural language queries for KPIs.
- **Category Filtering**: Access KPIs grouped by thematic areas.
- **Municipal & Regional Data Retrieval**: Fetch KPI data or historical time series.
- **Multi-Year Comparative Analysis**: Evaluate KPI performance changes across municipalities.
- **Cross-KPI Correlation**: Analyze relationships between KPIs.

## Available Tools
1. **list_operating_areas**: Retrieve available KPI categories.
2. **get_kpis_by_operating_area**: List KPIs under a category.
3. **search_kpis**: Discover KPIs using semantic search.
4. **get_kpi_metadata**: Access detailed KPI metadata.
5. **fetch_kolada_data**: Retrieve KPI values.
6. **analyze_kpi_across_municipalities**: In-depth municipal KPI analysis.
7. **compare_kpis**: Evaluate KPI correlations.
8. **list_municipalities**: List municipality IDs and names.

## Quick Start
Kolada MCP Server includes pre-cached KPI metadata. Delete `kpi_embeddings.npz` to refresh.

## Installation
Use `uv` to install Kolada MCP dependencies:

```bash
uv sync

Running Locally for Development

MCP Development (Claude Desktop)

Start the server locally:

uv run mcp dev server.py

Open MCP Inspector at http://localhost:5173 to test and debug.

HTTP Server (ChatGPT Integration)

Start the HTTP server for ChatGPT integration:

uv run kolada-mcp-http

The server will be available at http://localhost:8000. See for detailed ChatGPT integration instructions.

Claude Desktop Integration

Edit your claude_desktop_config.json to add Kolada MCP Server:

Docker Image (Local Build)

"KoladaDocker": {
  "args": [
    "run",
    "-i",
    "--rm",
    "--name",
    "kolada-mcp-managed",
    "kolada-mcp:local"
  ],
  "command": "docker",
  "env": {}
}

Prebuilt Container via PyPI

"KoladaPyPI": {
  "args": ["kolada-mcp"],
  "command": "/Users/hugi/.cargo/bin/uvx"
}

Local UV Execution (without Docker)

Replace [path to kolada-mcp] with your local directory:

"KoladaLocal": {
  "args": [
    "--directory",
    "[path to kolada-mcp]/src/kolada_mcp",
    "run",
    "kolada-mcp"
  ],
  "command": "uv"
}

Restart Claude Desktop after updating.

Contributing

Contributions are welcome! Submit issues, enhancements, or PRs on GitHub.

Disclaimer

Kolada MCP Server is independently developed, not endorsed by or affiliated with RKA or other organizations.

License

Kolada MCP Server is licensed under the .