Looker MCP. Run live queries directly against your data models.

You're tired of writing full SQL just to check one metric? This server connects your AI agent straight into Looker, letting it manage all your enterprise BI and data analytics through natural conversation. You don't need to touch the console or memorize IDs—you just talk to your agent, and it does the heavy lifting.

When you hook this up, your agent gets full control over the core functions of your Looker platform. It’s like having a dedicated BI analyst sitting right next to you, ready to pull any data point on demand without ever writing a manual query itself.

Running Live Queries and Debugging Data Logic

The most direct thing you can do is run ad-hoc queries against defined models. You use the run_inline_query tool to execute custom queries right away. This lets you fetch specific dimensions or measures in real time, giving you immediate answers without exporting data or writing complex SQL blocks.

If you need to know what's powering a report, you can investigate saved reports (Looks) using list_looks. This generates an inventory of all your Looks and shows their associated model mappings. You can then use the get_look tool for deep inspection; it traces the specific Looker report object ID to get detailed data maps and reveals exactly which filters were applied when that report was built.

It's critical for understanding data lineage.

Auditing and Inventorying Assets

Managing a big BI environment is a nightmare without good inventory tools, so we covered those too.

To figure out what dashboards exist, your agent first runs list_dashboards, pulling an instant list of every dashboard managed in the instance. If you need to know exactly how one specific dashboard works, you use get_dashboard. This tool retrieves not just the public details, but the complete underlying query structure for that dashboard ID.

It lets you audit the data logic without seeing a single line of SQL.

Need to find an asset buried deep in folders? You can run list_folders to enumerate the root folders and map out the high-level organizational structure across your entire tenant, helping you pinpoint where datasets are kept. For finding specific data assets by name or category, use search_content. This tool searches content metadata across the whole instance for key datasets or analytical materials.

What This Means For Your Workflow

You never have to manually navigate a dashboard list or search through folder trees. You simply ask your agent: 'What's the status of Q3 sales metrics?' It knows where to look, it pulls the list of relevant dashboards (list_dashboards), it checks if there's an existing report for that metric (list_looks), and then runs a targeted query using run_inline_query against the necessary model.

The agent synthesizes all this structured data and presents it back to you in plain text.

It’s about taking the guesswork out of your BI stack. You're not limited by whether or not someone created a dashboard for the specific question you have; you just ask, and your agent gives you the live data structure backing that answer. It keeps the whole process natural and fast.