home / mcp / claude code mcp server

Claude Code MCP Server

Modern Data Stack implementation with Meltano, DuckDB, and dbt Core for production-ready ELT pipelines

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "ajdoyl2-claude-data-stack-mcp": {
      "command": "./scripts/start_dbt_mcp.sh",
      "args": []
    }
  }
}

You have access to an MCP server that provides intelligent, context-aware assistance for dbt workflows. It exposes capabilities to run dbt CLI operations, discover project structure, query your DuckDB warehouse, and offer real-time development support, all to help you build, test, and transform your data stack more efficiently.

How to use

Use an MCP client to connect to the dbt MCP server. You can invoke dbt-related actions through the MCP to perform operations like running models, validating tests, exploring project metadata, and querying your analytics warehouse. The server exposes practical tools you can call in your development workflow to accelerate transformations and maintain data quality.

How to install

Prerequisites: you need a shell environment capable of running scripts and access to the project files that include the MCP server script. Ensure you have a compatible runtime for the server script you will execute.

# Start in a shell or terminal

# Ensure the script is executable
chmod +x ./scripts/start_dbt_mcp.sh

# Start the MCP server for dbt assistance
./scripts/start_dbt_mcp.sh

Configuration and usage notes

The MCP server is started via a dedicated script and does not require additional runtime steps to begin providing dbt assistance. It exposes a set of capabilities you can invoke from your MCP client, including dbt CLI operations, project discovery, and real-time guidance during development.

Tools and capabilities

dbt CLI Operations — Run, test, compile, and build dbt commands through the MCP to streamline your transformation workflow.

Project Discovery — List models, analyze metadata, and explore model lineage to understand how changes propagate through your analytics stack.

Database Querying — Execute SQL directly against your DuckDB warehouse to validate results and inspect data without leaving the development environment.

Real-time Assistance — Get context-aware guidance and practical workflows as you develop dbt projects.

Status and scope

Status: Production Ready with MCP Enhanced capabilities.

Available tools

dbt_run

Execute dbt run to compile and run models against the configured target, integrating with the MCP to show results and potential errors.

dbt_test

Run dbt tests to validate data quality and model expectations within your analytics project.

dbt_compile

Compile dbt models to verify syntax and model relationships without executing them.

dbt_build

Build dbt artifacts for documentation or deployment, enabling structure verification.

project_discovery

Discover models, sources, and metadata in the dbt project to understand lineage and dependencies.

lineage_exploration

Explore model lineage to see how changes impact downstream analytics and dashboards.

sql_query

Run direct SQL queries against DuckDB to inspect data and validate results in real time.