Usage

Video Demonstration

A demonstration of the MCP server is available here: Video Demo.

Common MCP Configuration

MCP clients that support an mcp.json configuration can start mcp_pyphotomol with uvx:

{
  "mcpServers": {
    "mcp_pyphotomol": {
      "command": "uvx",
      "args": ["mcp_pyphotomol"],
      "env": {
        "RESULTS_DIR": "/absolute/path/to/results-folder"
      }
    }
  }
}

RESULTS_DIR is the folder where plots and log files are stored. The server creates a date-stamped subfolder inside it for each run.

Claude Desktop

In Claude Desktop, open Settings, go to Developer, and click Edit Config. Add mcp_pyphotomol to claude_desktop_config.json:

{
  "mcpServers": {
    "mcp_pyphotomol": {
      "command": "uvx",
      "args": ["mcp_pyphotomol"],
      "env": {
        "RESULTS_DIR": "/Users/your-name/Documents/user_data_mcp_pyphotomol"
      }
    }
  }
}

Claude Desktop stores this file at:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Save the file, then fully quit and reopen Claude Desktop.

For local development, point the client at the repository checkout:

{
  "mcpServers": {
    "mcp_pyphotomol": {
      "command": "uvx",
      "args": [
        "--refresh",
        "--from",
        "/absolute/path/to/mcp_pyphotomol",
        "mcp_pyphotomol"
      ]
    }
  }
}

If you want to reuse the checkout’s existing environment, run it through uv:

{
  "mcpServers": {
    "mcp_pyphotomol": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/mcp_pyphotomol", "mcp_pyphotomol"]
    }
  }
}

ChatMCP Desktop

  1. Download the ChatMCP desktop app.

  2. Open the app and configure a model provider, such as OpenAI or Ollama.

  3. Verify that the chat is working by sending a message to the model.

  4. Open the settings and add a new MCP server.

  5. Set the server type to STDIO.

  6. For a local checkout, set the command to uv and the arguments to:

run --directory /absolute/path/to/mcp_pyphotomol mcp_pyphotomol

VS Code with GitHub Copilot

  1. Download and install Visual Studio Code.

  2. Set up GitHub Copilot in VS Code.

  3. Edit the VS Code MCP configuration, commonly named mcp.json:

{
  "servers": {
    "mcp_pyphotomol": {
      "command": "uvx",
      "args": ["mcp_pyphotomol"],
      "env": {
        "RESULTS_DIR": "/absolute/path/to/results-folder"
      }
    }
  }
}

  1. Start the MCP server from VS Code and use Copilot’s agent mode to interact with the tools.

Developer Debugging

To run the MCP server directly from a local checkout:

uv run mcp_pyphotomol

If your environment provides the FastMCP development CLI, you can also use it for interactive MCP debugging.