metabase-server MCP Server

A Model Context Protocol server for Metabase integration.

This is a TypeScript-based MCP server that implements integration with Metabase API. It allows AI assistants to interact with Metabase, providing access to:

• Detailed Metabase resources (databases, tables, fields, collections, etc.)
• A comprehensive suite of tools for listing, reading, executing, and managing Metabase assets.

Features

Resources

The server supports discovery and reading of a wide range of Metabase resources using metabase:// URIs. These resources can be listed hierarchically and read to get their full JSON representation.

• Databases: metabase://database/{db_id}
  • Includes details, list of tables, and schemas.
• Tables: metabase://database/{db_id}/table/{table_id}
  • Supports both regular tables and card-based models (e.g., metabase://database/{db_id}/table/card__{card_id}).
  • Includes table metadata and fields.
• Schemas: metabase://database/{db_id}/schema/{schema_name}
  • Lists tables within a specific schema.
• Collections: metabase://collection/{collection_id_or_root}
  • Supports listing items within a collection (cards, dashboards, sub-collections).
• Cards (Questions/Models): metabase://card/{card_id}
  • Detailed information about a specific question or model.
• Dashboards: metabase://dashboard/{dashboard_id}
  • Detailed information about a specific dashboard, including its dashcards and parameters.

Tools

The server provides a rich set of tools to interact with Metabase:

Listing & Reading Tools:

• list_dashboards: List Metabase dashboards.
  • Supports filtering by collection_id and archived status.
• list_cards: List Metabase cards (questions/models).
  • Supports filtering by collection_id, archived status, and f (filter type like 'archived', 'mine').
• list_databases: List Metabase databases.
  • Supports include_tables and include_cards to embed related resources.
• get_dashboard_details: Get full details for a specific dashboard, including its dashcards and parameters (dashboard_id).
• get_database_details: Get detailed information about a specific database (database_id), optionally including tables and fields.
• list_database_schemas: List all schemas within a specific database (database_id).
• list_tables_in_schema: List all tables within a specific schema of a database (database_id, schema_name).
• get_table_details: Get detailed metadata for a specific table (table_id) or a card-based model (card_model_id).
• get_field_distinct_values: Get distinct values for a specific field (field_id), useful for parameter population.
• list_collections: List Metabase collections.
  • Supports filtering by parent_collection_id and archived status.
• list_collection_items: List items (cards, dashboards, sub-collections) within a specific collection (collection_id).
  • Supports filtering by models (e.g., 'card', 'dashboard') and archived status.
• get_card_query_metadata: Get query metadata for a card (card_id), including dataset_query, result_metadata, and parameters.

Execution & Export Tools:

• execute_card: Execute a Metabase question/card (card_id) and get results.
  • Supports passing parameters as an array of objects.
• execute_query: Execute a native SQL query against a specified database (database_id, query).
  • Supports template_tags for substituting values in the SQL query.
• execute_dashboard_card_query: Execute a query for a card that is part of a dashboard (dashboard_id, dashcard_id, card_id), applying dashboard parameters.
• execute_mbql_query: Execute a query defined using Metabase Query Language (MBQL) (query object, optional parameters).
• export_query_results: Export results of a query (from card, native SQL, or mbql) in various formats (csv, xlsx, json, json_api).
  • Returns file content as base64 or structured JSON for json_api.

Parameter & Helper Tools:

• get_parameter_values: Fetches possible values for a parameter in a given context (context_type, context_id, parameter_id_or_slug).
  • Supports query for searching and constrained_parameters.

CRUD (Create, Read, Update, Delete) Tools:

• create_card, update_card, delete_card (supports hard/soft delete)
• create_dashboard, update_dashboard, delete_dashboard (supports hard/soft delete)

URI Examples

• Root (list databases and top-level collections): metabase://
• Database: metabase://database/1
• Schema within a database: metabase://database/1/schema/public
• Table within a database (default schema): metabase://database/1/table/23
• Table within a specific schema: metabase://database/1/schema/public/table/23 (Note: listing is via schema URI, reading table is direct)
• Card-based model: metabase://database/1/table/card__101 (Read via get_table_details or resource read)
• Collection: metabase://collection/5
• Root Collection: metabase://collection/root
• Card/Question: metabase://card/78
• Dashboard: metabase://dashboard/2

Configuration

Before running the server, you need to set the following environment variables:

# Required environment variables
export METABASE_URL=https://your-metabase-instance.com
export METABASE_USERNAME=your_username
export METABASE_PASSWORD=your_password

You can set these environment variables in your shell profile or use a .env file with a package like dotenv.

Development

Install dependencies:

npm install

Build the server:

npm run build

For development with auto-rebuild:

npm run watch

Installation

To use with Claude Desktop, add the server config:

On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json On Windows: %APPDATA%/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "metabase-server": {
      "command": "/path/to/metabase-server/build/index.js",
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_USERNAME": "your_username",
        "METABASE_PASSWORD": "your_password"
      }
    }
  }
}

Note: You can also set these environment variables in your system instead of in the config file if you prefer.

Installing via Smithery

To install metabase-server for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @traghav/metabase-server --client claude

Debugging

Since MCP servers communicate over stdio, debugging can be challenging. We recommend using the MCP Inspector, which is available as a package script:

npm run inspector

The Inspector will provide a URL to access debugging tools in your browser.