Usage

mcp_pykingenie is intended only for surface-based binding data, such as Octet and Gator BLI experiments.

Video Demonstrations

Two demonstration videos are available:

• Example with Visual Studio Code

• Example with Claude

Common MCP Configuration

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

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

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

Importing Data

Before importing data, ask the assistant to run print_data_dir if you want to use relative paths. Relative input paths are resolved inside that active date-stamped MCP data directory; absolute paths can be used directly.

The bundled Octet BLI example used by load_octet_example can be downloaded from the documentation as octet_bli_example_data.zip.

Example requests:

Load the Octet example experiment.
Import the Octet folder /Users/me/data/octet_run_01 as experiment "Run 01".
Import the Gator zip gator_run.zip from the MCP data directory as "Gator Run".
Import the KinGenie surface CSV /Users/me/data/surface_simulation.csv.

For Octet data, provide the folder containing the .frd files and sample plate metadata. For Gator data, provide either the folder or a .zip archive containing the channel CSV files plus Setting.ini and ExperimentStep.ini. For KinGenie surface CSV imports, provide a CSV with surface trace columns such as Time, Signal, Smax, and Analyte_concentration_micromolar_constant.

Claude Desktop

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

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

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

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

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

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_pykingenie mcp_pykingenie

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

4. 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_pykingenie

To use the HTTP transport:

uv run mcp_pykingenie -t http -p 8000

To use the MCP Inspector for interactive debugging against a local checkout:

npx @modelcontextprotocol/inspector uv --directory /Users/oburastero/Desktop/arise/mcp_pykingenie run mcp_pykingenie