Metabase

v0.1.0Analyticsstable

Metabase includes a built-in Model Context Protocol (MCP) server that lets AI clients connect directly to a Metabase instance. It uses the Streamable HTTP transport and builds on Metabase's Agent API to expose tools for searching, exploring, querying

analyticsbusiness-intelligenceclaudeclaude-codedata-analysis
Share:
76
Stars
0
Downloads
0
Weekly
0/5

What is Metabase?

Metabase is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to metabase includes a built-in model context protocol (mcp) server that lets ai clients connect directly to a metabase instance. it uses the streamable http transport and builds on metabase's agent api ...

Metabase includes a built-in Model Context Protocol (MCP) server that lets AI clients connect directly to a Metabase instance. It uses the Streamable HTTP transport and builds on Metabase's Agent API to expose tools for searching, exploring, querying

This server falls under the Analytics category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Metabase includes a built-in Model Context Protocol (MCP) se

Use Cases

Business intelligence and analytics
Data exploration and querying
Agent-driven BI tool access
jerichosequitin

Maintainer

LicenseMIT License
Languagetypescript
Versionv0.1.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y metabase

Manual Installation

npx -y metabase

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Metabase

The Metabase MCP server connects AI clients like Claude directly to a Metabase analytics instance, enabling natural-language querying, exploration, and export of your business intelligence data. It exposes five unified tools — list, retrieve, search, execute, and export — covering everything from browsing dashboards and cards to running custom SQL queries and exporting up to 1 million rows. With built-in response optimization (up to 90% token reduction), multi-layer caching, and read-only mode enforced by default, it is well-suited for both exploratory data analysis and production BI workflows.

Prerequisites

  • Node.js 18.0.0 or higher installed
  • A running Metabase instance (cloud or self-hosted) with a valid URL
  • A Metabase API key (recommended) or email/password credentials
  • An MCP-compatible client such as Claude Desktop or Claude Code
1

Obtain Metabase credentials

Log in to your Metabase instance and generate an API key under Settings > Admin > API Keys. Note your Metabase instance URL (e.g., https://analytics.example.com). API key authentication is recommended over email/password for security.

2

Add the server to your MCP client configuration

Open your MCP client config file (claude_desktop_config.json for Claude Desktop) and add the metabase-mcp server entry. Supply the required METABASE_URL and METABASE_API_KEY environment variables.

{
  "mcpServers": {
    "metabase-mcp": {
      "command": "npx",
      "args": ["-y", "@jerichosequitin/metabase-mcp"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_API_KEY": "your_api_key_here",
        "METABASE_READ_ONLY_MODE": "true",
        "EXPORT_DIRECTORY": "~/Downloads/Metabase",
        "CACHE_TTL_MS": "600000"
      }
    }
  }
}
3

Restart your MCP client

Save the configuration file and restart Claude Desktop (or reload the MCP server in your client). The metabase-mcp server will start automatically via npx on first use, downloading the package if needed.

4

Verify the connection

In your AI client, ask it to list available Metabase databases or dashboards. A successful response confirms the server is connected and authenticated.

5

(Optional) Enable write mode for SQL execution

By default, METABASE_READ_ONLY_MODE is true, blocking INSERT, UPDATE, DELETE, and DROP statements. Set it to false only if you need write access via the execute tool.

"METABASE_READ_ONLY_MODE": "false"

Metabase Examples

Client configuration

Full claude_desktop_config.json entry for Metabase MCP with API key authentication and default safe settings.

{
  "mcpServers": {
    "metabase-mcp": {
      "command": "npx",
      "args": ["-y", "@jerichosequitin/metabase-mcp"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_API_KEY": "your_api_key_here",
        "METABASE_READ_ONLY_MODE": "true",
        "EXPORT_DIRECTORY": "~/Downloads/Metabase",
        "CACHE_TTL_MS": "600000",
        "LOG_LEVEL": "info"
      }
    }
  }
}

Prompts to try

Example prompts that exercise the list, search, execute, and export tools.

- "List all dashboards in my Metabase instance."
- "Search for cards related to revenue in the Sales collection."
- "Run this SQL against database 2: SELECT date_trunc('month', created_at), COUNT(*) FROM orders GROUP BY 1 ORDER BY 1"
- "Retrieve the details of card ID 42 including its query and parameters."
- "Export the results of card 15 to CSV with the date filter set to last 30 days."

Troubleshooting Metabase

Authentication error: 401 Unauthorized

Verify that METABASE_URL does not have a trailing slash and that METABASE_API_KEY matches an active key in Metabase Admin > API Keys. If using email/password, set both METABASE_USER_EMAIL and METABASE_PASSWORD instead.

execute tool blocked with 'read-only mode' error

METABASE_READ_ONLY_MODE defaults to true, which blocks any non-SELECT SQL. Set it to false in your config env block to allow write queries, or use only SELECT statements.

Responses are slow or timeouts occur on large datasets

Increase REQUEST_TIMEOUT_MS (default 600000 ms) for very large queries, and use the export tool for datasets over 500 rows rather than the execute tool. Consider lowering CACHE_TTL_MS if cached data seems stale.

Frequently Asked Questions about Metabase

What is Metabase?

Metabase is a Model Context Protocol (MCP) server that metabase includes a built-in model context protocol (mcp) server that lets ai clients connect directly to a metabase instance. it uses the streamable http transport and builds on metabase's agent api to expose tools for searching, exploring, querying It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Metabase?

Install via npm with the command: npx -y metabase. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Metabase?

Metabase works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Metabase free to use?

Yes, Metabase is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Analytics MCP Servers

Explore all analytics servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "metabase": { "command": "npx", "args": ["-y", "metabase"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Metabase?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides