Skip to main content

​
Overview

Claude Desktop is Anthropic’s desktop AI assistant that supports MCP (Machine-Consumable Protocol) tools. Connect your Pylar tools to Claude Desktop to give Claude access to your data through governed views.

​
Prerequisites

  • ✅ Claude Desktop app installed on your computer
  • ✅ Pylar project with published MCP tools
  • ✅ MCP HTTP Stream URL and Bearer Token from Pylar

​
Step 1: Get Your Pylar Credentials

  1. In Pylar, navigate to your project
  2. Click “Publish” in the right sidebar
  3. Generate your token if needed
  4. Copy both:
    • MCP HTTP Stream URL: https://mcp.publish.pylar.ai/mcp
    • Authorization Bearer Token: Your unique token
Keep your Bearer Token secure. It provides access to your published tools.

​
Step 2: Open Claude Desktop Settings

  1. Open the Claude Desktop app
  2. Go to Settings → Developer
  3. Click “Edit Config”
This will open the claude_desktop_config.json file on your system.

​
Step 3: Add Pylar MCP Server Configuration

Add the following configuration to the claude_desktop_config.json file: 
{
  "mcpServers": {
    "view-mcp-node": {
      "url": "https://mcp.publish.pylar.ai/mcp",
      "headers": {
        "Authorization": "Bearer xxxxx"
      }
    }
  }
}
Important:
  • Replace xxxxx with your actual Bearer Token from Pylar
  • The name "view-mcp-node" is a local alias—you can rename it to anything you prefer
  • Save the file after making changes
If the “headers” key is unsupported in your Claude Desktop build, you may need to omit it and use a proxy or environment variable instead. Check your Claude Desktop version for MCP support.

​
Step 4: Restart Claude Desktop

  1. Save the claude_desktop_config.json file
  2. Restart Claude Desktop completely
  3. Claude Desktop will connect to your Pylar MCP server on startup

​
Step 5: Test the Connection

  1. Open a new conversation in Claude Desktop
  2. Ask Claude a question that requires using your MCP tool
Example: “Use the engagement score tool to get scores for login events” Claude will:
  1. Recognize it needs to use your Pylar MCP tool
  2. Call the tool with appropriate parameters
  3. Execute the query against your view
  4. Return the results and use them to answer your question
If Claude can successfully use your tools to answer questions, the connection is working correctly!

​
Using Your Tools

Once connected, you can ask Claude to use your Pylar tools in natural language:
  • “Get customer revenue data for the last quarter”
  • “Show me engagement scores for purchase events”
  • “What’s the average order value for customers in North America?”
Claude will automatically:
  • Identify which tool to use
  • Call the tool with the right parameters
  • Return formatted results
  • Use the data to provide comprehensive answers

​
Troubleshooting

​
Issue: Claude can’t see tools

Solutions:
  • Verify Bearer Token is correct (no extra spaces)
  • Ensure token format: Bearer YOUR_TOKEN (replace YOUR_TOKEN with actual token)
  • Check that tools are published in Pylar
  • Restart Claude Desktop completely
  • Verify the JSON syntax in the config file is valid

​
Issue: “Connection failed” error

Solutions:
  • Verify MCP HTTP Stream URL is correct: https://mcp.publish.pylar.ai/mcp
  • Check network connectivity
  • Ensure Bearer Token is valid and not expired
  • Regenerate token in Pylar if needed
  • Check Claude Desktop version supports MCP headers

​
Issue: Tools return errors

Solutions:
  • Check Evals in Pylar to see detailed error information
  • Verify view queries are correct
  • Test tools in Pylar before using in Claude
  • Review tool parameters and ensure they’re being passed correctly

​
Issue: Headers not supported

If your Claude Desktop build doesn’t support headers in the config: Alternative Solutions:
  • Use a proxy server to add authentication headers
  • Set authentication via environment variables (if supported)
  • Check for Claude Desktop updates that add header support

​
Benefits

By connecting Claude Desktop to Pylar:
  • ✅ Secure Access: Claude only accesses data through your governed views
  • ✅ Natural Language: Ask Claude questions in plain English
  • ✅ Centralized Control: Update views in Pylar, Claude gets updates automatically
  • ✅ Full Visibility: Monitor Claude’s usage through Pylar Evals
  • ✅ No Code: No need to write integration code
  • ✅ Real-time Updates: Changes to views and tools reflect immediately

​
Next Steps

Monitor Usage

See how Claude interacts with your tools through Evals