Skip to content
Looker MCP Server

Looker MCP Server

by Google
Added to Onei
3 views
Repository Docs
Databases & Data

Summary

Google's MCP server built into the Looker platform: agents query Explores and dashboards through Looker's governed semantic layer over OAuth 2.1, with no middleware to run.

Features

  • MCP endpoint built into the Looker platform — no middleware or proxy to deploy
  • Queries resolve through the LookML semantic layer, so Looker generates the SQL
  • Inherits Looker's access controls; the agent acts as the authenticated user
  • Admin-controlled tool allowlist governs exactly which capabilities are exposed
  • OAuth 2.1 with PKCE; clients registered via the oauth_client_apps API

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Looker MCP Server" tools will be available.
{
  "mcpServers": {
    "looker-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://YOUR-LOOKER-INSTANCE/mcp"
      ]
    }
  }
}

Description

The Looker-managed MCP server embeds a Model Context Protocol endpoint directly in the Looker platform, letting Gemini CLI, Claude Desktop, Cursor and any other MCP client reach an organisation's business metrics without a separate proxy or gateway to deploy.

The useful property is that queries go through Looker's semantic layer rather than around it. An agent asks for a metric; LookML resolves what that metric means and generates the SQL. This avoids the usual failure mode of pointing a model at a warehouse and hoping it writes correct joins, and it means every answer inherits Looker's existing access controls — row-level permissions, model access and user attributes all still apply, because the agent acts as the authenticated user rather than through a shared service account.

Capabilities include running queries against Explores, retrieving dashboards, checking instance health, and pulling live business metrics. The exact tool list is not fixed: administrators control it through a tool allowlist in the Admin panel's MCP page, so an organisation can expose read-only querying without exposing anything operational.

The endpoint lives at <your-looker-instance>/mcp and authenticates with OAuth 2.1 using PKCE. Setup has one manual step worth planning for — an admin must register each AI client as an OAuth application through the oauth_client_apps API or the API Explorer before it can connect.

This is distinct from connecting to Looker through the open-source MCP Toolbox for Databases, which remains the route for customer-hosted instances. The managed server is in preview and requires a Looker-hosted instance, either Looker (Google Cloud core) or Looker (original); customer-hosted deployments are not yet supported.

Related MCP Servers

MCP: AntV Chart MCP

by AntV

New

Official AntV MCP server that turns a description of your data into a rendered chart — 26 visualization types from bar and line to sankey, mind-map, fishbone and geographic maps.

Developer ToolsDatabases & Data
1 views

MCP: InfluxDB 3 MCP Server

by InfluxData

Featured

InfluxData's official MCP server for InfluxDB 3 — schema discovery, bounded SQL and InfluxQL queries, line-protocol writes and token administration across Core, Enterprise and Cloud.

Databases & Data
3 views
Browse all MCP servers →