Home / MCP / Systems MCP Server

Systems MCP Server

Provides an MCP server to run lethain:systems models and prime context with documentation and examples.

python

15stars

View on GitHub

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
    "mcpServers": {
        "systems": {
            "command": "uv",
            "args": [
                "--directory",
                "/Users/your_user/systems-mcp",
                "run",
                "main.py"
            ]
        }
    }
}

You can run and prime systems models locally using the MCP server named systems. It exposes two practical tools for modeling with lethain:systems and for priming context to improve model quality. This guide walks you through how to use the server with an MCP client, how to install it locally, and important notes to help you get productive quickly.

How to use

Use an MCP client to interact with the systems MCP server. The server exposes two tools you can call directly from your client:

run_systems_model: Executes the systems specification of a systems model. You provide the specification and, optionally, the number of rounds to run (default is 100).
load_systems_documentation: Loads documentation and examples into the context window to prime models for more helpful systems modeling.

How to install

Prerequisites and setup to run locally require a Python-ready environment, a compatible MCP runtime (UV-based or similar), and a client configured to connect to the MCP server.

Follow these concrete steps to install and configure the MCP server for local use with Claude Desktop on macOS.

1. Install Claude Desktop.
2. Clone systems-mcp into a convenient location, for example:
   /Users/your_user/systems-mcp
3. Ensure you have uv installed and available in your PATH.
4. Open Claude Desktop, go to Settings > Developer, and allow it to create your MCP config file.
5. Update your claude_desktop_config.json to include the MCP server configuration:

   {
     "mcpServers": {
       "systems": {
         "command": "uv",
         "args": [
           "--directory",
           "/Users/your_user/systems-mcp",
           "run",
           "main.py"
         ]
       }
     }
   }

6. Close Claude and reopen it. The MCP server should be available for use.

Additional notes

If you want to prime the context further, you can run the documentation-loading tool to inject examples and guidelines into the session. This can help the model generate more accurate system specifications.

Available tools

run_systems_model

Runs the systems specification of a systems model. Accepts the specification and an optional number of rounds (default 100).

load_systems_documentation

Loads documentation and examples into the context window to prime models for more helpful systems modeling.