tecton-mcp

Tecton MCP Server

This is a Mission Control Protocol (MCP) server from Anthropic for Tecton that provides a set of tools to interact with Tecton clusters, manage feature stores, and execute Tecton CLI commands.

Features

The server provides the following MCP tools:

CLI Tools

• tecton_cli_help: Get structured help information about available Tecton CLI commands
• tecton_cli_execute: Execute Tecton CLI commands

Feature Store Management

• list_workspaces: List all workspaces in the connected Tecton cluster
• list_feature_views: List all feature views with their metadata
• list_feature_services: List all feature services with their metadata
• list_transformations: List all transformations with their metadata
• list_data_sources: List all data sources with their metadata
• list_entities: List all entities with their metadata

Configuration Tools

• get_feature_service_configuration: Get detailed configuration of a feature service
• get_feature_view_configuration: Get detailed configuration of a feature view
• get_feature_view_code: Get the Python code definition of a feature view

Setup

Prerequisites

• Python >=3.10 or compatible version
• Tecton SDK installed and configured
• Mission Control Protocol (MCP) installed

Installation

1. Install required Python packages:

pip install httpx click cloudpickle

2. Install Tecton SDK:

pip install tecton

3. Install MCP:

pip install mcp

Configuration

Add the following to your MCP server configuration:

{
    "mcpServers": {
        "tecton": {
            "command": "/path/to/python",
            "args": [
                "--directory",
                "/path/to/tecton",
                "run",
                "tecton.py"
            ],
            "env": {
                "PYENV_VERSION": "3.9.11"
            }
        }
    }
}

Replace /path/to/python and /path/to/tecton with your actual paths.

Usage

Starting the Server

1. First, ensure you have Tecton configured and logged in:

tecton login

2. Then run the server using:

python tecton.py

The server will start and listen for MCP commands.

Using the Tools

All tools are available through the MCP interface. Here are some example uses:

1. List all workspaces:

workspaces = await list_workspaces()

2. Get feature view configuration:

config = await get_feature_view_configuration(name="my_feature_view", workspace="my_workspace")

3. Execute a Tecton CLI command:

result = await tecton_cli_execute(command="workspace list")

Error Handling

The server includes comprehensive error handling:

• All tools return empty lists or empty strings on failure
• Errors are logged using the _err function
• General operations are logged using the _log function

Dependencies

• Core Python:

  • typing (built-in)
  • httpx
  • click
  • cloudpickle

• Tecton:

  • tecton
  • tecton._internals
  • tecton.cli.cli
  • tecton_core
  • tecton_proto

• MCP:

  • mcp.server.fastmcp

• Local:

  • utils (containing _err, _log, and run_command)

Contributing

Feel free to submit issues and enhancement requests!