Log inSubmit server
Metabase AI Assistant logo

Metabase AI Assistant

Connects AI clients to Metabase for SQL work, dashboards, dbt metadata, lineage analysis, and governed data operations.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Metabase AI Assistant

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "metabase-ai-assistant": {
      "command": "npx",
      "args": [
        "-y",
        "metabase-ai-assistant"
      ],
      "env": {
        "METABASE_URL": "YOUR_VALUE_HERE",
        "METABASE_API_KEY": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Automated check passed— started and listed 134 tools correctly (2d ago).

Overview

The metabase-ai-assistant MCP server connects MCP-compatible AI clients to a Metabase instance through its API. It provides tools for SQL generation and execution, dashboard creation, database inspection, dbt metadata and lineage workflows, semantic governance, anomaly detection, and query optimization. Run it with NPX or install it from npm, then configure the Metabase URL and API key. Reach for it when an AI assistant needs to query, analyze, or manage Metabase-backed analytics with optional read-only controls.

Use cases

  • Generate and repair SQL against Metabase-connected databases
  • Build dashboards from dbt or BI metadata
  • Inspect dbt models and resolve multi-hop lineage
  • Govern semantic business rules with approval states
  • Detect anomalies and review query index recommendations

Key features

  • Metabase SQL generation and execution
  • Self-healing SQL retries
  • Dashboard construction from YAML metadata
  • dbt metadata and lineage analysis
  • Semantic rule approval workflows
  • PII and token masking

Tools134

What Metabase AI Assistant lets your agent do. Expand a tool for its parameters.

Verified live~15.7k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 3d ago

Recent health check succeeded.

7/7 checks healthy over the last 40d

What it can do

From the 134 tools this server returned when we ran it; 134 declare their own safety hints.

Runs commands · 5Can delete or modify data · 12Makes network requests · 3Read-only tools · 117
Network at startup3 connections
Files written at startupNone
Install scriptsNone
npm provenanceNot published
Package age269 days
Maintainers1

Observed 2026-10-20 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 15d ago via OSV.dev · metabase-ai-assistant (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars49Stargazers on the source repository.
npm downloads203Package downloads in the last 30 days.
Last commit1mo agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed134Callable tools this server registers over MCP.
Directory activity2 viewsConfig copies, upvotes, and views on AllMCPs.

How Metabase AI Assistant works

What metabase-ai-assistant MCP server does

The metabase-ai-assistant MCP server exposes Metabase operations to language models, coding assistants, and automated workflows through MCP. Its tools cover SQL generation and execution, database introspection, dashboard construction, query analysis, anomaly detection, and index recommendations. It also includes workflows for working with dbt metadata, MetricFlow semantic models, lineage, and governed business rules.

The project supports Metabase Open Source and Enterprise deployments and documents compatibility from Metabase 0.43 through current 0.61+ releases. It can detect or work across different Metabase API and query formats, including session-token and API-key authentication paths.

How it works

The metabase-ai-assistant MCP server runs as a Node.js process and communicates with an MCP client over the configured transport. A client sends tool calls, and the server uses the Metabase API and related local metadata inputs to perform the requested operation.

The SQL workflow includes ai_sql_execute_and_heal, which can retry a failed query through an automated recovery loop. Documented recovery cases include syntax errors, misspelled columns, missing grouping clauses, and dialect differences across PostgreSQL, MySQL, BigQuery, Snowflake, and SQLite. Fix history is retained in a healing trail.

For dbt-oriented work, tools can inspect manifests and semantic models, prioritize sources, resolve multi-hop lineage, advise on pre-aggregations, build dashboards from YAML metadata, and export approved semantic definitions to YAML. Semantic memory operations use proposal, approval, deprecation, restoration, and listing stages rather than silently changing or deleting rules.

Setup and configuration

Install with NPX:

Terminal
npx metabase-ai-assistant

Alternatively, install the package globally with npm. Configure the MCP client to launch the command and provide these environment variables:

  • METABASE_URL: URL of the Metabase deployment.
  • METABASE_API_KEY: API key used to authenticate with Metabase.
  • METABASE_READ_ONLY_MODE: optional setting shown in the example client configuration for restricting operations.

The README provides configurations for Claude Desktop, Cursor, Windsurf, and VS Code. It also documents an SSE/HTTP mode that exposes an OpenAPI document for ChatGPT Actions, plus deployment instructions for Cloudflare Workers.

Tools and capabilities

The metabase-ai-assistant MCP server includes tools and workflows for:

  • Generating, executing, and repairing SQL.
  • Inspecting databases, models, and Metabase metadata.
  • Creating and arranging dashboards from structured definitions.
  • Detecting anomalies and advising on query indexes.
  • Scanning dbt projects and inspecting MetricFlow models.
  • Resolving lineage and suggesting pre-aggregation designs.
  • Proposing, approving, deprecating, restoring, and listing semantic rules.
  • Masking emails, phone numbers, national identifiers, payment-card numbers, IP addresses, and tokens in output.

Limitations and notes

The documentation presents different tool totals in different sections, including 143, 152, and the repository listing’s 111. Treat the exact count as version-dependent or unresolved rather than relying on one figure. Read-only mode is shown as a client configuration option, but write-capable operations remain available unless that setting is used. Database dialect behavior and Metabase features may vary by deployment version and configuration.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Metabase AI Assistant

metabase-ai-assistant is an MCP server that connects AI clients to Metabase through its API. Its main tools support SQL generation and execution, self-healing query retries, dashboard creation, database inspection, dbt metadata and lineage analysis, semantic rule governance, anomaly detection, and query index recommendations.

Install it with `npx metabase-ai-assistant` or run `npm install -g metabase-ai-assistant`. Configure an MCP client to launch the package, then provide `METABASE_URL` and `METABASE_API_KEY` in the client environment.

Yes, metabase-ai-assistant documents Cursor configuration through `.cursor/mcp.json`. Add the server command and provide the Metabase URL, API key, and optional read-only setting in the environment section.

Yes, it includes dashboard-building capabilities, including a tool that translates dbt and Lightdash-style YAML metadata into Metabase dashboard layouts. The documentation also describes end-to-end dashboard architecture workflows.

Yes, the client configuration examples set `METABASE_READ_ONLY_MODE` to `true`. This provides a documented way to request read-only operation when connecting the server to an MCP client.

Related MCP servers

All in Databases All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewMetabase AI Assistant AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/metabase-ai-assistant?style=directory)](https://allmcps.com/mcp/metabase-ai-assistant)
HTML Embed
<a href="https://allmcps.com/mcp/metabase-ai-assistant"><img src="https://allmcps.com/api/badge/metabase-ai-assistant?style=directory" alt="Metabase AI Assistant on AllMCPs" /></a>