How to Connect Tableau to Gemini via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Tableau, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Tableau + Gemini

Work with Tableau Cloud and Tableau Server from any AI agent: find workbooks, views and data sources, read view data as CSV, images, PDFs and Excel crosstabs, query published data sources through the VizQL Data Service, trace lineage with the Metadata API, and manage projects, tags, group members and extract refreshes. Sign-in with a personal access token.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

TABLEAU_SERVER_URL, TABLEAU_PAT_NAME, TABLEAU_PAT_SECRET (optional: TABLEAU_SITE)

Step 1 — Get credentials

Setting up takes about 5 minutes. You need a Tableau Cloud site or Tableau Server 2024.2 or later, and a Tableau user who can see the content the AI should work with.

1. In Tableau open your Account Settings, section Personal Access Tokens, enter a token name and click Create Token. On Tableau Server an admin may have to allow personal access tokens first.
2. Copy the token name into TABLEAU_PAT_NAME and the secret, which is shown only once, into TABLEAU_PAT_SECRET. Both are case-sensitive.
3. Put the address without a path into TABLEAU_SERVER_URL: on Tableau Cloud the pod host you see in the browser, e.g. https://prod-useast-a.online.tableau.com, on Tableau Server its own address, e.g. https://tableau.example.com.
4. Put the site's content URL into TABLEAU_SITE: the part after /site/ in the browser address (#/site/acme/home gives acme). Leave it empty for the Default site of a Tableau Server.
5. Install the connector and run tableau_get_current_session: it shows the site and the user and confirms the sign-in. Its site.id is the site_id every site tool needs.

The connector signs in with the token, reuses the session and signs in again when it expires. Tableau allows one session per token at a time, so do not use the same token in another tool. Tableau Cloud revokes a token that has not been used for 15 days; create a new one then.

Whose access it is: the AI sees and changes only what the token's user may see and do, with that user's site role and permissions.

What it reads: the server version and newest REST API version, the sites and their settings. A word search across workbooks, views, data sources, projects, flows, metrics, tables and databases. Projects, workbooks with their views and data connections, and views with usage counts. What a dashboard shows: the data behind a view as CSV, a PNG image the model can look at, a PDF of one view or of a whole workbook, and the crosstab as an Excel file, each narrowed with view filters if needed. Published data sources with connections, certification and extract information, their fields (tableau_read_datasource_metadata) and queries on them with dimensions, aggregated measures, filters and sorting (tableau_query_datasource). Lineage through GraphQL on the Metadata API (tableau_metadata_query): which workbooks use a data source, upstream databases, tables and columns, calculated fields, owners and data quality warnings. Users, groups and group members, background jobs and scheduled extract refresh tasks.

What it writes: refreshes the extracts of a workbook or a published data source now, runs a scheduled extract refresh task outside its schedule, creates and updates projects, renames a workbook or switches its tabs on or off, moves a workbook to another project (its permissions may follow the target project's settings), adds tags to workbooks and data sources, and adds a user of the site to a local group (Active Directory groups sync from the directory). A refresh returns a job that the AI follows with tableau_get_job.

Switched off at install: tableau_cancel_job, because a cancelled job cannot be resumed and has to be started again. An admin switches it on when the team needs it.

Data formats: ids are LUIDs (UUIDs). Lists come as {"pagination": {"pageNumber", "pageSize", "totalAvailable"}, "workbooks": {"workbook": [...]}}, and the same for projects, views, data sources, users, groups and jobs. Times are ISO 8601 in UTC. tableau_get_view_data returns CSV text with a header row, with numbers formatted as in the view. tableau_query_datasource returns data as a list of objects keyed by field caption with the function, e.g. SUM(Sales). View filters use the vf_ prefix and the field name as shown in the view, e.g. {"vf_Region": "West"}, with several values comma-separated. A job's finishCode is 0 for success, 1 for failure and 2 for cancelled.

Paging, filters and sorting: list tools take page_size (up to 1000) and page_number; the AI continues while pageNumber * pageSize is below totalAvailable. tableau_search_content pages with limit and page, and its answer carries the total. filter is field:operator:value, several joined by commas, e.g. projectName:eq:Finance,updatedAt:gt:2026-01-01T00:00:00Z; in takes [a,b]. sort is field:asc or field:desc.

Requirements: tableau_read_datasource_metadata and tableau_query_datasource need Tableau Cloud or Tableau Server 2025.1 with the VizQL Data Service on, and the API Access capability on the data source. tableau_metadata_query needs the Metadata API, which is on by default on Tableau Cloud and is enabled by an admin on Tableau Server. tableau_search_content needs Tableau Cloud or Tableau Server 2022.3 and later.

Common errors: Tableau answers {"error": {"code", "summary", "detail"}}. 401 (codes 401001 and 401002) means a wrong or expired token or a wrong site content URL. 403 means the token's user lacks the permission. 404 means a wrong id or a site_id from another site; at sign-in it means TABLEAU_SERVER_URL is wrong (only scheme and host) or the server is older than 2024.2. 400 is a bad filter or parameter, 409 a duplicate name or a refresh that is already queued.

The connector does not publish, download or delete workbooks and data sources, does not delete projects, does not add users to the site or change site roles, and does not remove tags or group members.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Gemini

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Gemini CLI reads MCP servers from ~/.gemini/settings.json (on Windows %USERPROFILE%\.gemini\settings.json). Add:

{
  "mcpServers": {
    "anythingmcp": {
      "httpUrl": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart gemini.
3. Run /mcp inside Gemini CLI (or gemini mcp list in your shell): anythingmcp should show as connected.
4. Start prompting: Gemini CLI can now call the Tableau tools.

Available tools

Example prompts

• "Which workbooks in the Finance project were updated in the last 30 days, and who owns them?"
• "Show me the numbers behind the Sales Overview dashboard, filtered to the West region."
• "In the Superstore data source, what were total sales by region in the first half of 2026?"
• "Which workbooks use the published data source Orders? I want to know what breaks if we change it."
• "Did last night's extract refreshes succeed? Show me the failed ones with their error text."

FAQ

Where does Gemini CLI look for MCP servers? In ~/.gemini/settings.json for your user, or in .gemini/settings.json inside a project. gemini mcp list shows each server and whether it connects.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT
• Connect to Cursor
• Connect to Meta Muse
• Connect to Microsoft Copilot