Tableau MCP Server

A powerful Model Context Protocol (MCP) server that connects Claude Desktop to Tableau Server, enabling natural language interactions with your Tableau data and comprehensive administrative capabilities.

- List & Browse: Explore all workbooks, views, and data sources
- Data Extraction: Get CSV data from any Tableau view
- Visual Export: Download dashboard images in PNG format
- Complete Datasets: Download entire data sources with all raw data
- Smart Search: Find content across your Tableau server

- User Management: List all users with roles and login history
- Permission Auditing: See exactly who has access to workbooks
- Group Management: View all groups and member counts
- Project Oversight: Audit project permissions and settings
- Site Administration: Get comprehensive site statistics
- Usage Analytics: Detailed workbook usage and access audits

- "Show me the sales data from Q4 dashboard"
- "Export the customer metrics as CSV"
- "Find all reports containing revenue data"

- "Who has access to the Finance workbook?"
- "List all users who haven't logged in recently"
- "Show me all groups and their member counts"
- "Audit the permissions for our HR dashboards"

- Python 3.8+
- Claude Desktop installed
- Access to Tableau Server or Tableau Cloud
- Administrative privileges (for admin tools)

# Clone the repository git clone https://github.com/hetpatel-11/Tableau-MCP.git cd Tableau-MCP # Create virtual environment (recommended) python -m venv tableau_mcp_env source tableau_mcp_env/bin/activate # On Windows: tableau_mcp_env\Scripts\activate # Install ALL dependencies including MCP SDK pip install -r requirements.txt

2. MCP SDK Installation (If Having Issues)

If you encounter MCP SDK installation issues, try these solutions:

# Option 1: Install MCP SDK directly pip install mcp>=1.0.0 # Option 2: Install with specific version pip install mcp==1.0.0 # Option 3: Install from source (if needed) pip install git+https://github.com/modelcontextprotocol/python-sdk.git # Option 4: Force reinstall pip install --force-reinstall mcp>=1.0.0

# Test that MCP SDK is properly installed python -c "import mcp; print('✅ MCP SDK installed successfully')" # Test Tableau Server Client python -c "import tableauserverclient as TSC; print('✅ Tableau Server Client ready')" # Test the server python tableau_mcp_server.py --test

Create a.envfile with your Tableau credentials:

TABLEAU_SERVER_URL=https://your-tableau-server.com TABLEAU_SITE_ID=your-site-name # Option 1: Personal Access Token (Recommended) TABLEAU_TOKEN_NAME=your-token-name TABLEAU_TOKEN_VALUE=your-token-value # Option 2: Username/Password # TABLEAU_USERNAME=your-username # TABLEAU_PASSWORD=your-password

Add to your Claude Desktop configuration:

macOS:~/Library/Application Support/Claude/claude_desktop_config.json

{ "mcpServers": { "tableau-server": { "command": "/path/to/Tableau-MCP/start_clean.sh", "args": [] } } }

# Make startup script executable chmod +x start_clean.sh # Start the server ./start_clean.sh

The server exposes Tableau content as MCP resources:

- tableau://workbooks/{workbook_id}- Workbook metadata
- tableau://views/{view_id}/data- View data in CSV format

Claude Desktop ↔ MCP Server ↔ Tableau REST API ↔ Tableau Server ⬇️ [15 Available Tools] ⬇️ [Data + Admin Capabilities]

Tableau-MCP/ ├── tableau_mcp_server.py # Main MCP server implementation ├── start_clean.sh # Server startup script ├── requirements.txt # Python dependencies (includes MCP SDK) ├── setup.py # Setup and configuration script ├── env.example # Environment configuration template ├── README.md # This file └── .env # Your credentials (create this)

- Local Operation: Server runs locally with your permissions
- Secure Credentials: Uses environment variables for authentication
- Token Authentication: Supports Tableau Personal Access Tokens
- No Data Leakage: No data sent to external services

"ModuleNotFoundError: No module named 'mcp'"

# Solution 1: Install MCP SDK pip install mcp>=1.0.0 # Solution 2: Check virtual environment source tableau_mcp_env/bin/activate pip install -r requirements.txt # Solution 3: Verify Python version python --version # Should be 3.8+

# Uninstall and reinstall pip uninstall mcp pip install mcp>=1.0.0

- Verify credentials in.envfile
- Check Tableau Server URL format
- Ensure account has proper permissions

"No tools available in Claude"

- Restart Claude Desktop after configuration
- Check thatstart_clean.shis executable
- Verify Python dependencies are installed

- Ensure your account has administrative privileges
- Check Tableau Server user permissions
- Verify site access rights
- Test Connection:python tableau_mcp_server.py --test
- Debug Mode:python tableau_mcp_server.py --debug
- Check Logs: Look for error messages in terminal output
- Verify MCP:python -c "import mcp; print('MCP SDK OK')"

User: "Show me all workbooks and find the sales data" Claude: [Lists workbooks] → [Finds sales-related content] → [Extracts data] User: "Get the customer metrics from the Q4 dashboard as CSV" Claude: [Locates dashboard] → [Extracts view data] → [Provides CSV]

User: "Who has access to our Finance workbook?" Claude: [Lists all users and groups with permissions] → [Shows permission levels] User: "Show me all users who are site administrators" Claude: [Lists users by role] → [Highlights administrators]

- Fork the repository
- Create a feature branch
- Add your improvements
- Test thoroughly
- Submit a pull request

This project is licensed under the MIT License - see theLICENSEfile for details.

- Built with theModel Context Protocol
- UsesTableau Server Client
- Designed forClaude Desktop

A read-only MCP server for SAP BusinessObjects BI, powered by the CData JDBC Driver.