dbt Conversation AI Local

What it does

Connects an LLM to a local dbt project via the Model Context Protocol, allowing users to query semantic layers, run models, and explore metrics using natural language without data leaving the machine.

Tools

• build: Execute dbt build to materialize models.
• compile: Compile dbt models to SQL.
• docs: Generate dbt documentation.
• list: List dbt models, seeds, and snapshots.
• parse: Parse the dbt project to generate the manifest.
• run: Run specific dbt models.
• test: Execute dbt data tests.
• show: Preview dbt model results.
• metricflow_list_metrics: Retrieve a list of available metrics.
• metricflow_list_dimensions: List dimensions for the semantic layer.
• metricflow_list_semantic_models: List defined semantic models.
• metricflow_query: Query the semantic layer for specific metrics and dimensions.

Installation

{
  "mcpServers": {
    "dbt-conversation-ai": {
      "command": "uv",
      "args": ["run", "dbt-mcp"]
    }
  }
}

Supported hosts

• Claude Desktop