CargoEffe Server

Connect AI assistants (Claude Desktop, Claude Code, Cursor) to CargoEffe for automated cargo loading plan design.

Installation

# Option A: pip install
pip install cargoeffe-mcp

# Option B: uvx (no pre-install needed)
uvx cargoeffe-mcp

# Option C: from source
git clone https://github.com/dollob/cargoeffe-mcp.git
cd cargoeffe-mcp && pip install -e .

Setup

1. Get your MCP token from CargoEffe Settings → MCP Tokens
2. Configure your AI client:

Claude Code

{
  "mcpServers": {
    "cargoeffe": {
      "command": "uvx",
      "args": ["cargoeffe-mcp"],
      "env": {
        "CARGOEFFE_MCP_TOKEN": "cfm_YOUR_TOKEN_HERE"
      }
    }
  }
}

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "cargoeffe": {
      "command": "python",
      "args": ["-m", "cargoeffe_mcp.server"],
      "env": {
        "CARGOEFFE_MCP_TOKEN": "cfm_YOUR_TOKEN_HERE"
      }
    }
  }
}

Debug Mode

Set CARGOEFFE_MCP_DEBUG=true for verbose logging:

{
  "env": {
    "CARGOEFFE_MCP_TOKEN": "cfm_...",
    "CARGOEFFE_MCP_DEBUG": "true"
  }
}

Debug logs appear on stderr (visible in Claude Code's MCP debug panel).

Available Tools

Security

This package is a thin protocol bridge — it contains NO business logic, database queries, or API secrets. All sensitive operations happen on the CargoEffe backend.

License

MIT