Tableau MCP
Official Tableau MCP server: the agent finds data sources, sends queries to them and opens dashboards and Pulse metrics
- 361 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Some tools change data: they publish sources, delete content, run flows and edit refresh tasks
- Local mode acts with the rights of the PAT owner
- Datasource queries hand business data from Tableau to the agent
Install
- Works in
Manual install
{
"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"
}
}
}
}The local setup from the README; needs Node.js 22.7.5 or later. Replace the env values with your own.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Some tools change data: they publish sources, delete content, run flows and edit refresh tasks
- Local mode acts with the rights of the PAT owner
- Datasource queries hand business data from Tableau to the agent
This is third-party code. Review the repository files before installing.
What it does
Tableau MCP connects an agent to your Tableau Cloud or Tableau Server site. The agent finds workbooks, views and data sources, reads field metadata and sends queries to the data without leaving the chat. For example, it can name the top five states by sales in the Superstore data source, find the most viewed workbook of the year or show the Economy view from the Finances project. For questions about data the server offers data source search and listing, metadata reading and a datasource query. For Pulse metrics it builds short insight briefs and lists definitions, values and subscriptions. For administrators it runs flows, lists jobs, edits extract refresh schedules and queries Admin Insights. Separate tools publish a data source, delete content and scaffold a data app. There are two ways to run it. For Tableau Cloud, Tableau hosts a managed service at https://mcp.tableau.com: each user signs in with OAuth 2.1 under their own account, and their Tableau permissions apply as usual. Locally or on your own server it runs through npx, Docker or from source and signs in to Tableau with a personal access token (PAT).
Who it is for. For Tableau analysts and administrators who want to query published data and manage site content through an agent.
Good fit when
- You want to ask a published data source such as Superstore for the five states with the most sales in 2025
- You need the most viewed workbook of the year with its owner and project
- The agent should show a specific view, for example «Economy» from the «Finances» project
- You want a summary of a Pulse metric or a list of metric subscriptions
- An administrator needs to check jobs and extract refresh tasks on the site
Not a fit when
- You have no Tableau Cloud or Server site: the server works only through their API
- You want to build dashboards from scratch in the visual editor: the server reads data and manages content, it does not replace the Tableau interface
- The agent should not be allowed to delete or run flows: the server has a delete-content tool, so narrow that access with tool scoping or PAT permissions
- The site belongs to a client and you have no right to issue an access token for it
Example request
In the Superstore data source, find the five states with the biggest sales in 2025 and show how far the first leads the secondMCP
- Transport
- stdio, http
- Authentication
- API key
Environment variables
SERVERrequired- Address of your Tableau Server or Tableau Cloud
SITE_NAMErequired- Tableau site name
PAT_NAMErequired- Name of the personal access token
PAT_VALUErequired, secret- Secret value of the personal access token
Limitations
Local setup needs Node.js 22.7.5 or later and a Tableau personal access token. The token is stored in the client's environment variables, so issue a separate narrow-scoped token just for the agent.
The managed service covers Tableau Cloud only. Tableau Server owners and anyone who needs their own infrastructure have to self-host the server using the deployment guides in the documentation. The Docker image has to be rebuilt after every repository update.
Tableau itself supports the project, as the Tableau Supported badge shows. Whether Tableau Cloud is reachable from Russia has not been checked.
How to disable. Remove the tableau block from mcpServers in your client settings and restart it. Revoke the PAT you issued for the agent in your Tableau account settings.
FAQ
How do I connect Tableau MCP to Claude?
For Tableau Cloud the managed service is simpler: add its address to your client as a remote MCP server and complete the OAuth sign-in. Step-by-step instructions for Slack, Claude, ChatGPT and other clients are on the Popular Client Integrations page of the documentation. For your own server, add an npx launch block with four environment variables.
Which variables does the npx setup need?
Four: SERVER with your Tableau address, SITE_NAME with the site name, PAT_NAME with the personal access token name and PAT_VALUE with its secret. PAT_NAME takes the token's name; a username or email will not work there.
What access does the agent get?
In the managed service each user signs in under their own account and keeps their Tableau permissions. In local mode the agent acts with the rights of the PAT owner, so narrow the token's permissions to the projects you need.
Can the server run in Docker?
Yes. The getting-started guide shows how to build the image with npm run build:docker and configure the client with docker run, the --env-file flag and an env.list file.
Is there a template for cloud deployment?
Yes, the repository has a Deploy to Heroku button with app.json and a Procfile, and the documentation has a page on experimental Heroku support. Enterprise setups for Tableau Cloud and Tableau Server have their own sections.
Discussion
No questions or reviews yet.
Related
MCP serversA local Microsoft MCP server for Power BI semantic models: an agent creates tables, DAX measures and relationships in Desktop, Fabric and PBIP files
Official MongoDB MCP server: queries and aggregations, index management, Atlas clusters and local deployments
A desktop SQL workspace with a built-in MCP server: the agent reads the schema and runs queries in the same client you already use
The official MathWorks MCP server: the agent starts and quits MATLAB, writes and runs code, and checks it for errors and style
MCP server and CLI to drive real Microsoft Excel: Power Query, DAX, VBA, PivotTables and charts
Official Apache IoTDB MCP server: SQL over time series in tree and table models, with CSV and Excel export
README in short
Getting Started opens with the managed service for Tableau Cloud and a sample npx config. Below are a Heroku deploy button and the server's answers to three typical questions.