AllTrails

You can run an MCP server that exposes trail data through a standardized Model Context Protocol interface. This guide walks you through practical usage, installation steps, and essential notes, so you can run the server locally and connect from an MCP client. The project described here is historical and provided for educational purposes; the server is deprecated and should not be used against live AllTrails data.

How to use

You interact with the AllTrails MCP Server by connecting your MCP client to the server using the standard MCP protocol over stdio. Start the server, then issue natural-language style requests like: search for trails in a specific park, or fetch detailed information about a named trail. The server exposes two core functions: searching trails by park and retrieving detailed trail data such as difficulty, length, elevation gain, and route summaries. Use the client to ask for trails in a park, and then request details for a chosen trail.

How to install

Prerequisites: Python 3.8 or higher. A MCP-compatible client to connect to the server.

Option 1: With Virtual Environment (Recommended) to isolate dependencies.

git clone <your-repo-url>
cd alltrails_mcp

python3 -m venv alltrails_mcp_venv
source alltrails_mcp_venv/bin/activate  # On Windows: alltrails_mcp_venv\Scripts\activate

pip install -r requirements.txt

python3 server.py

Option 2: With System Python if you prefer not to use a virtual environment.

git clone <your-repo-url>
cd alltrails_mcp

pip install -r requirements.txt

python3 server.py

Option 3: With Docker to run in a container.

docker build -t alltrails-mcp .

docker run -i --rm alltrails-mcp

After starting, you can configure your MCP client to connect to the local stdio server as described in the configuration section.

Additional sections

Configuration and client setup details are shown below. The server communicates via standard input/output (stdio) and uses MCP protocol version 1.9.4.

Claude Desktop users can configure the MCP server to run the local Python process. The example below demonstrates how to point Claude Desktop at a local Python script that starts the server.

{
  "mcpServers": {
    "alltrails_mcp_server": {
      "command": "python3",
      "args": ["server.py"]
    }
  }
}

Find your Claude Desktop config file if you need to customize the MCP server connection. Typical paths include the user profile's application support directory on macOS or the AppData directory on Windows.

Troubleshooting tips include checking server startup, verifying Python environment paths, and ensuring all dependencies are installed. If you encounter a connection issue, test the server manually by running the Python script in a terminal, then observe logs for errors.

Common command to verify the server starts without errors is to run the server script directly, then stop with Ctrl+C when finished.

Available tools

search_trails

Search for trails in a specified national park using trail data, returning matching trail results with basic metadata.

get_trail_details

Retrieve detailed information about a specific trail by its AllTrails slug, including metrics like difficulty, length, and elevation.