The Model Context Protocol (MCP) allows external AI assistants like Claude Desktop to connect to your Kubiya platform and execute workflows.

​
What is MCP?

MCP is an open standard that lets AI assistants connect to external data sources and tools. When you connect Claude Desktop to Kubiya via MCP, Claude can:
  • Browse your workflows and agents
  • Execute workflows on your behalf
  • Query your connected systems
  • Access your knowledge bases

​
Quick Setup

  1. Install the Kubiya MCP Server 
    npm install -g @kubiya/mcp-server
  2. Configure Claude Desktop Add to your Claude Desktop configuration: 
    {
  "mcpServers": {
    "kubiya": {
      "command": "kubiya-mcp-server",
      "env": {
        "KUBIYA_API_KEY": "your-api-key",
        "KUBIYA_ORG": "your-org"
      }
    }
  }
}
  3. Restart Claude Desktop Claude will now have access to your Kubiya workflows and can execute them through natural conversation.

​
Usage Example

Once connected, you can ask Claude:
“Deploy the frontend service to staging using the deployment workflow”
Claude will:
  1. Find your deployment workflow
  2. Ask for any required parameters
  3. Execute the workflow through Kubiya
  4. Show you the results

​
Security

  • MCP connections use your existing Kubiya API keys
  • All executions go through Kubiya’s normal security controls
  • You can revoke access anytime by removing the MCP server configuration

​
Troubleshooting

Connection Issues: Verify your API key and organization name are correct Permission Errors: Ensure your API key has the necessary permissions for the workflows you want to execute Workflow Not Found: Check that the workflow name matches exactly (case-sensitive)