Calgary Open Data MCP Server

An MCP (Model Context Protocol) server that gives AI assistants access to the Calgary Open Data portal. Search datasets, inspect schemas, and query any dataset using SoQL — all through a unified MCP interface.

Disclaimer: This project is unofficial and is not affiliated with, endorsed by, or connected to the City of Calgary. It queries publicly available data through the Socrata Open Data API.

Quick Start

Use the hosted server (no setup required)

Connect any MCP-compatible client to the hosted endpoint:

https://calgary-open-data.baobabtech.app/mcp

Run locally with npx

npx calgary-open-data-mcp

This starts a local stdio MCP server — no cloning or configuration needed.

Related MCP server: CKAN MCP Server

Tools

The server exposes three tools:

search_datasets

Search the Calgary Open Data catalog by keyword, category, or tag.

get_dataset_metadata

Get the full schema and metadata for a specific dataset.

Returns column names, field names, data types, descriptions, tags, license, and last updated timestamp.

query_dataset

Query any dataset using SoQL (Socrata Query Language).

Client Configuration

All examples below use the hosted endpoint. Replace the URL with http://localhost:3000/mcp for local development.

Claude Desktop

Settings > Connectors > Add custom connector, enter:

https://calgary-open-data.baobabtech.app/mcp

Claude Code

claude mcp add --transport http calgary-open-data https://calgary-open-data.baobabtech.app/mcp

Cursor

In .cursor/mcp.json:

{
  "mcpServers": {
    "calgary-open-data": {
      "url": "https://calgary-open-data.baobabtech.app/mcp"
    }
  }
}

ChatGPT

Settings > Apps & Connectors > Advanced settings, enable Developer mode, click Create, enter:

https://calgary-open-data.baobabtech.app/mcp

npx (stdio) for any MCP client

{
  "mcpServers": {
    "calgary-open-data": {
      "command": "npx",
      "args": ["-y", "calgary-open-data-mcp"]
    }
  }
}

Self-Hosting

Vercel

The root of this repo is a Next.js app ready to deploy on Vercel.

git clone https://github.com/baobab-tech/calgary-open-data-mcp.git
cd calgary-open-data-mcp
pnpm install

Deploy via the Vercel CLI or connect the repo in the Vercel dashboard.

Optional: Set a SOCRATA_APP_TOKEN environment variable in your Vercel project settings for higher Socrata API rate limits. Register for a free app token at data.calgary.ca.

Cloudflare Workers

A self-contained Cloudflare Worker lives in the cloudflare/ directory.

cd cloudflare
pnpm install
pnpm run login       # authenticate with Cloudflare (first time only)
pnpm run dev         # local development
pnpm run cf:deploy   # deploy to Cloudflare

Optional: Set a SOCRATA_APP_TOKEN secret:

npx wrangler secret put SOCRATA_APP_TOKEN

Development

Prerequisites

• Node.js >= 22

• pnpm

Setup

git clone https://github.com/baobab-tech/calgary-open-data-mcp.git
cd calgary-open-data-mcp
pnpm install

Running locally

Vercel (Next.js) dev server:

pnpm dev

MCP endpoint available at http://localhost:3000/mcp.

Cloudflare Worker dev server:

cd cloudflare
pnpm install
pnpm run dev

CLI (stdio):

pnpm build:cli
node dist/cli.js

Type checking

# Root project (Vercel + CLI)
pnpm tsc --noEmit

# Cloudflare Worker
cd cloudflare && pnpm tsc --noEmit

Building the CLI for npm

pnpm build:cli

This produces dist/cli.js — the file referenced by the bin field in package.json.

Project Structure

src/
  lib/
    socrata.ts              Socrata API client (Discovery, Metadata, SODA)
    tools.ts                Shared MCP tool definitions
  cli.ts                    npx entry point (stdio transport)
  app/
    [transport]/route.ts    Vercel HTTP endpoint (Streamable HTTP)
    page.tsx                Landing page
    layout.tsx
cloudflare/
  src/index.ts              Cloudflare Worker (self-contained)
  wrangler.toml
  package.json

The Vercel deployment and the npx CLI share tool definitions via src/lib/tools.ts. The Cloudflare Worker is self-contained since CF Workers use a different module runtime.

How It Works

This server wraps three layers of the Socrata Open Data API that powers data.calgary.ca:

1. Discovery API — catalog search across all datasets on the portal

2. Metadata API — detailed schema and column information per dataset

3. SODA API — the data query layer, using SoQL (a SQL-like query language)

Every dataset on the portal has a unique four-by-four ID (e.g. 35ra-9556). Use search_datasets to find IDs, get_dataset_metadata to understand the schema, then query_dataset to pull the data.

Transport

All remote endpoints use Streamable HTTP — the current MCP transport standard. The npx CLI uses stdio transport for local use.

Contributing

Contributions are welcome! Here's how to get started:

1. Fork the repository

2. Create a feature branch (git checkout -b feature/my-feature)

3. Make your changes

4. Run type checking (pnpm tsc --noEmit)

5. Commit your changes (git commit -m 'Add my feature')

6. Push to the branch (git push origin feature/my-feature)

7. Open a Pull Request

Ideas for contributions

• Additional tools (e.g. dataset export, geospatial queries)

• Support for other Socrata-powered open data portals

• Caching layer for frequently accessed metadata

• Better error messages and input validation

• Tests

Please open an issue first if you're planning a large change so we can discuss the approach.

License

MIT

Acknowledgements

• City of Calgary Open Data for making public data accessible

• Socrata / Tyler Data & Insights for the API platform

• Model Context Protocol for the open standard