Content
# Toronto Open Data MCP Server
An MCP (Model Context Protocol) server that provides direct access to Toronto's Open Data through the CKAN API. This server allows LLM agents to efficiently discover, explore, and query Toronto's 500+ public datasets.
## Features
- 🔍 **Smart Dataset Discovery**: Search through 500+ Toronto datasets with intelligent suggestions
- 🧠 **Intelligent Data Helper**: Automatically handles both API and CSV data sources
- 📊 **Flexible Querying**: Support for filtering, sorting, and field selection on API datasets
- 📁 **CSV Support**: Automatic fetching and preview of downloadable CSV datasets
- 🤝 **LLM-Friendly**: Designed for collaborative use with web search when additional context is needed
- ✅ **Robust Error Handling**: Clear error messages with actionable suggestions
## Installation
### Option 1: Using uvx (Recommended for MCP)
The easiest way to use this server with any MCP client:
```bash
brew install uv
```
```bash
# No installation needed! uvx will handle everything
# Just use in your MCP client configuration:
uvx toronto-open-data-mcp-server
```
### Option 2: Development Installation
1. **Clone the repository**:
```bash
git clone https://github.com/yourusername/toronto-open-data-mcp-server.git
cd toronto-open-data-mcp-server
```
2. **Install dependencies**:
```bash
# Install main dependencies
pip install -e .
# Install test dependencies (optional)
pip install -e ".[test]"
```
3. **Run the server**:
```bash
python main.py
```
## MCP Client Configuration
### Using uvx (Easiest Method)
Add this to your MCP client configuration:
```json
{
"mcpServers": {
"Toronto Open Data Server": {
"command": "uvx",
"args": ["toronto-open-data-mcp-server"]
}
}
}
```
### Configuration File Locations
- **Claude Desktop**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Cursor**: Check Cursor's MCP documentation for config location
- **Other MCP clients**: Refer to your client's documentation
## Usage
### Quick Start
The server provides several tools designed for LLM agents:
1. **`toronto_start_here()`** - Essential first call that explains the workflow
2. **`toronto_search_datasets(query)`** - Find relevant datasets by keywords
3. **`toronto_smart_data_helper(dataset_id, user_question)`** - Intelligent data retrieval
4. **`toronto_popular_datasets()`** - Quick access to commonly used datasets
### Example Workflow
```python
# 1. Start with guidance
toronto_start_here()
# 2. Search for relevant data
toronto_search_datasets("restaurant inspection")
# 3. Get data intelligently
toronto_smart_data_helper("dinesafe", "recent restaurant inspection failures")
# 4. Advanced filtering (if needed)
toronto_query_dataset_data("dinesafe",
filters={"establishment_status": "Conditional Pass"},
sort="inspection_date desc",
limit=10)
```
### Popular Datasets
- **`dinesafe`** - Restaurant inspections and health scores
- **`traffic-signals`** - Traffic light locations and timing
- **`parks-facilities`** - Parks, pools, and recreation facilities
- **`business-licences`** - Licensed businesses in Toronto
- **`building-permits`** - Construction and renovation permits
## Testing
This project includes comprehensive tests covering unit tests, integration tests, and workflow tests.
### Prerequisites
Install test dependencies:
```bash
pip install -e ".[test]"
```
### Running Tests
#### Quick Test Commands
```bash
# Run unit tests only (recommended for development)
python run_tests.py
# Run with verbose output
python run_tests.py --verbose
# Run with coverage report
python run_tests.py --coverage
# Run integration tests (hits real Toronto API)
python run_tests.py --integration
# Run all tests (unit + integration)
python run_tests.py --all
```
## API Reference
### Core Tools
#### `toronto_start_here() -> str`
Essential first call that provides workflow guidance and server capabilities.
#### `toronto_search_datasets(query: str, limit: int = 10) -> str`
Search Toronto datasets by keywords.
#### `toronto_smart_data_helper(dataset_id: str, user_question: str, limit: int = 10) -> str`
Intelligent helper that automatically handles both API and CSV data sources.
#### `toronto_query_dataset_data(dataset_id: str, filters: Dict = None, fields: List = None, limit: int = 10, sort: str = None) -> str`
Advanced querying with filtering and sorting for API datasets.
### Utility Tools
#### `toronto_popular_datasets() -> str`
Quick access to commonly used datasets.
#### `toronto_get_dataset_schema(dataset_id: str) -> str`
Get field names and types for API datasets.
#### `toronto_fetch_csv_data(csv_url: str, max_lines: int = 50) -> str`
Fetch and preview CSV file content.
## Architecture
- **FastMCP Framework**: Built on the FastMCP framework for easy tool definition
- **CKAN API**: Direct integration with Toronto's CKAN-based Open Data portal
- **Collaborative Design**: Works alongside web search rather than replacing it
- **Error Recovery**: Intelligent error handling with actionable suggestions
## Contributing
1. Fork the repository
2. Create a feature branch
3. Add tests for new functionality
4. Run the test suite: `python run_tests.py --all`
5. Submit a pull request
