Tableau MCP

Overview

Tableau MCP is a suite of developer primitives, including tools, resources and prompts, that will make it easier for developers to build AI applications that integrate with Tableau.

Official Documentation

https://tableau.github.io/tableau-mcp/

Quick Start

Requirements

• Node.js 22.7.5 or newer
• An MCP client e.g. Claude Desktop, Cursor, VS Code, MCP Inspector, etc.

Standard config works in most MCP clients:

{
  "mcpServers": {
    "tableau": {
      "command": "npx",
      "args": ["-y", "@tableau/mcp-server@latest"],
      "env": {
        "SERVER": "https://my-tableau-server.com",
        "SITE_NAME": "my_site",
        "PAT_NAME": "my_pat",
        "PAT_VALUE": "pat_value"
      }
    }
  }
}

Optional: Database Proxy Integration

If you expose a POST endpoint that runs parametrized SQL (for example the Bun HTTP proxy in server/ at http://127.0.0.1:8787/db/query), set DB_PROXY_URL before starting the server:

export DB_PROXY_URL=http://127.0.0.1:8787/db/query
npm run start:http

This enables the query-db MCP tool. To store reusable procedures and refresh rules, also set BUN_SERVER_BASE_URL to the Bun API origin (defaults to http://127.0.0.1:8787).

Prompt examples:

• “Use the query-db tool to fetch the last 10 orders from the sales_dashboard table.”
• “Query the Postgres database (not Tableau) for total sales by region.”
• “Store a query expectation for sales_dashboard with ID latest-orders so the monitor can reuse it.”
• “If total sales exceed 50,000, use the dashboard status API to set updateRequired:true.”
• “Configure the dashboard monitor so it refreshes when sales_dashboard total sales is above 50,000.”
• “Simulate December 2028 sales for the West region using the forecast tool.”

query-db accepts { "query": "SELECT ...", "params": [] } and forwards the request to the proxy URL. store-query-expectation accepts { "id": "...", "query": "...", "expectedValue": "{\"foo\":true}" } and persists the payload via POST /db/expectations. configure-dashboard-monitor accepts { "threshold": 50000, "dashboardId": "...", "tableName": "..." } and calls POST /dashboard/monitor. simulate-sales accepts { "month": "December", "year": 2028, "filterColumn": "region", "filterValue": "West" } and internally uses query-db to build a projected dataframe. It also stores the detailed rows via POST /simulations so dashboards can retrieve the simulated dataset.

Deploy to Heroku

Example Prompts to Client

Prompt #1: Querying Data

For the Superstore Datasource, what are the top 5 states with the most sales in 2025?

Response

Based on the Superstore Datasource, here are the top 5 states with the most sales in 2025:
  1. California - $146,388.34
  2. New York - $93,922.99
  3. Washington - $65,539.90
  4. Texas - $43,421.76
  5. Pennsylvania - $42,688.31
California leads by a significant margin with over $146,000 in sales, followed by New York with nearly $94,000. These two states alone account for a substantial portion of the total sales across the top 5 states.

Prompt #2: Content Exploration

Find me the most viewed workbook within the last year.

Response

The most viewed workbook within the last year is:

Global Connections
- Owner: John Doe
- Project: Marketing
- Views in last year: 523
- Last Modified: September 2, 2025
- Workbook ID: 111fb672-9401-4913-a167-17b3d89a3e4f

Prompt #3: Getting a View

Show me an image of the "Economy" view in the "Finances" project.

Response