Looker MCP Server

by Google / Looker

MCP server that connects AI agents to Looker’s semantic layer for trusted, governed data access

See Google Cloud Looker MCP documentation

Summary

The Looker MCP Server provides a secure, model-agnostic bridge between AI applications and enterprise data by exposing Looker’s semantic layer through the Model Context Protocol (MCP). Agents can request metrics and explorers in natural language and receive grounded SQL/query results from Looker without hand-coding SQL.

Features

  • Model-agnostic MCP interface to Looker’s semantic layer
  • Generates optimized SQL from natural language requests using Looker’s governed definitions
  • Works with Gemini CLI, Claude Desktop, Cursor, and other MCP clients
  • Inherits Looker security, access controls, and audit logging
  • Supports creation of Looks and dashboards via agent workflows

Superpowers

Enables product teams and analysts to query trusted enterprise metrics with AI agents while preserving governance and auditability.

Known limitations & notes

  • Requires Looker and configured data connections (BigQuery, AlloyDB, Cloud SQL, etc.)
  • Admins must configure appropriate permissions for agent access

Sources / notes:

  • Google Cloud Looker documentation and product announcements.