Metabase MCP Server

A Model Context Protocol (MCP) server for Metabase that gives AI assistants full access to your analytics platform — dashboards, cards, databases, tables, collections, and more.

Developed and maintained by EaseCloud — cloud-native, AI-driven, and data infrastructure solutions.

Quick Start

export METABASE_URL=https://your-metabase-instance.com
export METABASE_API_KEY=your_metabase_api_key
npx @easecloudio/mcp-metabase-server

96 Tools Available

Supported Metabase Versions

• Metabase v0.46.x and above (recommended: v0.48.x or later)
• Metabase Cloud (fully supported)
• Self-hosted instances (Docker, JAR, or cloud deployments)

Installation

npx (Recommended)

npx @easecloudio/mcp-metabase-server

Global install

npm install -g @easecloudio/mcp-metabase-server
mcp-metabase-server

Docker

docker build -t mcp-metabase-server .
docker run -it --rm \
  -e METABASE_URL=https://your-metabase-instance.com \
  -e METABASE_API_KEY=your_metabase_api_key \
  mcp-metabase-server

Configuration

Authentication

API Key (preferred):

METABASE_URL=https://your-metabase-instance.com
METABASE_API_KEY=your_metabase_api_key

Username / Password (fallback):

METABASE_URL=https://your-metabase-instance.com
METABASE_USERNAME=your_username
METABASE_PASSWORD=your_password

Copy .env.example to .env and fill in your values.

Tool Filtering (TOOL_MODE)

Set TOOL_MODE to control which tools are exposed to the AI. Useful for limiting surface area or preventing accidental writes.

TOOL_MODE=essential  # smallest surface area
TOOL_MODE=read       # read-only
TOOL_MODE=all        # everything (default)

Claude Desktop Integration

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

Using npx:

{
  "mcpServers": {
    "metabase": {
      "command": "npx",
      "args": ["@easecloudio/mcp-metabase-server"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_API_KEY": "your_metabase_api_key"
      }
    }
  }
}

With tool filtering:

{
  "mcpServers": {
    "metabase": {
      "command": "npx",
      "args": ["@easecloudio/mcp-metabase-server"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_API_KEY": "your_metabase_api_key",
        "TOOL_MODE": "essential"
      }
    }
  }
}

Using a local build:

{
  "mcpServers": {
    "metabase": {
      "command": "node",
      "args": ["/path/to/metabase-server/dist/index.js"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_API_KEY": "your_metabase_api_key"
      }
    }
  }
}

Username / password fallback:

{
  "mcpServers": {
    "metabase": {
      "command": "npx",
      "args": ["@easecloudio/mcp-metabase-server"],
      "env": {
        "METABASE_URL": "https://your-metabase-instance.com",
        "METABASE_USERNAME": "your_username",
        "METABASE_PASSWORD": "your_password"
      }
    }
  }
}

Available Tools

Core CRUD

Card Layout

Public Sharing & Embedding

Revision, Audit & Discovery

Core CRUD

Execution & Export

Parameters

Metadata & Discovery

Public Sharing & Embedding

Core CRUD

Querying

Schema & Sync

Diagnostics

Metadata

Card Virtual Tables

Management

CSV Upload (Metabase-managed tables)

Collections

Users

Permissions

Search

These tools enable an AI to convert native SQL questions into interactive MBQL questions. Metabase has no REST API for this conversion — it requires field IDs, which these tools cache locally.

How it works:

1. get_card — extract the SQL and database_id
2. get_schema_cache — get tables + field IDs for MBQL references like ["field", 42, null]
3. AI translates SQL → MBQL using the cached field IDs
4. create_card — save the new interactive question

Cache is stored at ~/.easecloud/metabase-mcp/cache/{url-hash}/ with a 24-hour TTL and scoped per Metabase instance.

MCP Resources

Access Metabase entities directly via metabase:// URIs:

Development

npm install
npm run build       # compile TypeScript → dist/
npm run watch       # incremental rebuild
npm run dev         # build + start
npm run inspector   # launch MCP Inspector for debugging

Debugging

MCP servers communicate over stdio, which makes direct debugging awkward. Use the MCP Inspector:

npm run inspector

The Inspector provides a browser UI for sending tool calls and inspecting responses.

About EaseCloud

EaseCloud is a cloud consulting and solutions company specializing in:

• Cloud-native application development
• AI & automation integrations
• DevOps and infrastructure management
• Data analytics and BI platform consulting

We built this project to contribute to the open-source MCP ecosystem while demonstrating our expertise in integration, automation, and cloud solutions.

If your team is adopting Metabase at scale or looking to integrate AI with your BI stack, get in touch — we provide consulting, customization, and managed support for enterprises.

📧 [email protected] 🌐 easecloud.io

Bug Reports & Issues

Found a bug or have a feature request?

🐛 Create a GitHub issue

Please include:

• Metabase version
• MCP server version
• Steps to reproduce
• Expected vs actual behavior
• Any error messages or logs

Contributing

Contributions are welcome. Visit the GitHub repository to submit issues or pull requests.

License

MIT — see LICENSE for details.