dbt-mcp

Overview

The dbt MCP server is a production-ready implementation by dbt Labs that enables AI agents to interact with dbt projects through the Model Context Protocol. It offers extensive functionality across multiple categories including SQL execution, semantic layer operations, dbt CLI commands, and documentation management. The server provides tools for model discovery, lineage analysis, metric queries, and even code generation, making it a powerful integration for data teams looking to enhance their dbt workflows with AI assistance.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• execute_sql

  Executes SQL on dbt Platform infrastructure with Semantic Layer support.

  Try:Run a SQL query to see top 10 customers by revenueTry:Execute a SQL query against the production database
• text_to_sql

  Generates SQL from natural language using project context.

  Try:Convert 'show me monthly sales for each product' to SQLTry:Generate SQL for total revenue by region
• query_metrics

  Executes metric queries with filtering and grouping options.

  Try:Query monthly revenue metricsTry:Get metrics filtered by product category
• get_model_details

  Gets model details including compiled SQL, columns, and schema.

  Try:Show details for the customers modelTry:Get information about the orders model
• get_lineage

  Gets full lineage graph with type and depth filtering.

  Try:Show the lineage for the revenue modelTry:Get upstream and downstream dependencies
• run

  Executes models to materialize them in the database.

  Try:Run all models in the projectTry:Execute the staging models
• generate_model_yaml

  Generates model YAML with columns, option to inherit upstream descriptions.

  Try:Generate YAML for a new customer modelTry:Create model YAML with columns from upstream
• fusion.compile_sql

  Compiles SQL in project context via dbt Platform.

  Try:Compile this SQL with dbt context: SELECT * FROM customersTry:Help me understand what this SQL would compile to
• search_product_docs

  Searches docs.getdbt.com for pages matching a query.

  Try:Find documentation about incremental modelsTry:Search for information on dbt tests
• get_job_details

  Gets job configuration including triggers, schedule, and dbt commands.

  Try:Show details for the nightly jobTry:Get configuration for the weekly refresh
• list_metrics

  Retrieves all defined metrics.

  Try:Show all metrics in the projectTry:List all revenue-related metrics
• get_model_health

  Gets health signals: run status, test results, and upstream source freshness.

  Try:Check the health status of the customers modelTry:Show test results and freshness for all models

Comparable tools

Installation

{
  "mcpServers": {
    "dbt": {
      "command": "npx",
      "args": ["@dbt-labs/dbt-mcp"],
      "env": {
        "DBT_PROJECT_DIR": "/path/to/your/dbt/project"
      }
    }
  }
}

npm install -g @modelcontextprotocol/mcpb
mcpb install https://github.com/dbt-labs/dbt-mcp/releases/download/vX.X.X/dbt-mcp.mcpb

FAQ

What is the dbt MCP server?

The dbt MCP server is a Model Context Protocol implementation that provides AI agents with tools to interact with dbt projects, including SQL execution, semantic layer operations, and dbt CLI commands.

Which dbt products does this server support?

The server supports dbt Core, dbt Fusion, and dbt Platform, with varying functionality available depending on your specific dbt setup and access level.

On Hacker News

Recent discussion from the developer community.

• MCP Server for Working with Dbt▲ 1
  Story by duck · 2025-04-21

Compare dbt-mcp with