arcade-mcp-camino

Camino MCP Server

An MCP (Model Context Protocol) server for the Camino AI API, built with the Arcade MCP Server framework. This server provides AI agents with location intelligence, spatial reasoning, and route planning capabilities.

Features

This MCP server exposes all Camino API endpoints as tools:

• search_place - Search for places using free-form or structured queries
• query - Natural language queries converted to Overpass QL for POI discovery
• spatial_relationship - Calculate distance, direction, and travel time between points
• place_context - Get context-aware information about locations (nearby places, weather, etc.)
• journey_planner - Multi-waypoint journey planning with route optimization
• get_route - Detailed routing between two points with geometry and imagery

Installation

Prerequisites

• Windows users: Must use WSL (Windows Subsystem for Linux) - see CLAUDE.md for details
• Python 3.10+
• uv package manager
• A Camino API key

Setup

1. Clone the repository:

git clone https://github.com/yourusername/arcade-mcp-camino.git
cd arcade-mcp-camino/camino_server

2. Install dependencies:

uv sync

3. Create a .env file with your Camino API key:

CAMINO_API_KEY=your-api-key-here

Usage

With Claude Desktop (Windows via WSL)

Add to your Claude Desktop config (%APPDATA%\Claude\claude_desktop_config.json):

{
  "mcpServers": {
    "camino_server": {
      "command": "wsl",
      "args": [
        "bash", "-c",
        "cd /mnt/c/Users/YOUR_USERNAME/path/to/arcade-mcp-camino/camino_server && uv run python -m camino_server.server"
      ],
      "env": {
        "CAMINO_API_KEY": "your-api-key-here"
      }
    }
  }
}

With Claude Desktop (macOS/Linux)

{
  "mcpServers": {
    "camino_server": {
      "command": "uv",
      "args": [
        "run",
        "python",
        "-m",
        "camino_server.server"
      ],
      "cwd": "/path/to/arcade-mcp-camino/camino_server",
      "env": {
        "CAMINO_API_KEY": "your-api-key-here"
      }
    }
  }
}

Standalone

Run the server directly:

uv run python -m camino_server.server

Example Usage

Once configured in Claude Desktop, you can ask questions like:

• "Find coffee shops near the Eiffel Tower"
• "What's the distance and travel time from Times Square to Central Park?"
• "Plan a journey from my hotel to the museum, then to a restaurant"
• "Get me context about what's around coordinates 40.7589, -73.9851"
• "Find the route from San Francisco to Los Angeles"

Performance Notes

• The query endpoint may take 30-60 seconds as it processes complex Overpass API queries and AI ranking
• Other endpoints typically respond in 1-5 seconds
• Use mode="basic" for faster, free queries or mode="advanced" for enhanced web-enriched results

Troubleshooting

See CLAUDE.md for common issues and solutions, including:

• Windows fcntl module errors
• Return type validation errors
• Timeout issues with the query endpoint

API Documentation

Full API documentation available at: https://getcamino.ai/docs

License

MIT