Registering an MCP server

In the Cloudera console, click the Cloudera AI tile. The Cloudera AI Workbenches page is displayed.
Click on the name of the workbench. The workbench Home page is displayed.
In the left navigation pane, click AI Studios.
Click the Open button on the Agent Studio tile. The Agent Studio window is displayed.
Go to Tools Catalog in the Agent Studio window.
Go to the MCP Servers tab.
Click the Register button. The MCP server registration modal window is displayed.

Figure 1. Register MCP Server modal window
Provide the MCP server configuration in the following format:
```
{
    "mcpServers": {
        "<ServerName>": {
            "command": "<runtime>",
            "args": ["<package-name>"],
            "env": {
                "VARIABLE_NAME": "value"
            }
        }
    }
}
```
note
Agent Studio uses the same configuration format as Claude Desktop for MCP server registration.
- The following example shows a configuration for a Python-based MCP server using the uvx command-line tool:
```
{
    "mcpServers": {
        "DuckDuckGo": {
            "command": "uvx",
            "args": ["duckduckgo-mcp-server"]
        }
    }
}
```
- The following example shows a configuration for a Node.js-based MCP server using the npx command-line tool:
```
{
    "mcpServers": {
        "Puppeteer": {
            "command": "npx",
            "args": ["-y", "@modelcontextprotocol/server-puppeteer"]
        }
    }
}
```
- The following example shows a configuration for an MCP server with environment variables:
```
{
  "mcpServers": {
    "awslabs.amazon-sns-sqs-mcp-server": {
      "command": "uvx",
      "args": ["awslabs.amazon-sns-sqs-mcp-server@latest"],
      "env": {
        "AWS_PROFILE": "your-aws-profile",
        "AWS_REGION": "us-east-1"
      }
    }
  }
}
```
- The following example shows a configuration with placeholders:
```
{
    "mcpServers": {
        "DatabaseMCP": {
            "command": "uvx",
            "args": ["database-mcp-server"],
            "env": {
                "DATABASE_URL": "postgresql://user:password@host:port/database",
                "API_KEY": "your-api-key-here"
            }
        }
    }
}
```
note

For security reasons, the Agent Studio backend does not store environment variable values, so you must use placeholder values during registration and provide the actual credentials when configuring the server within a workflow.

Following the registration, Agent Studio performs the following actions:

Validates the MCP server setup.
Identifies available tools by reflecting on the server.
Catalogs these tools for subsequent use within workflows.

Figure 2. MCP Servers tab in Agent Studio