One-Click Deploy
Navigate to the Server page and click Deploy. This triggers:- Python source code is generated from all configured components
- The FastMCP server process starts inside the MCP container
- The server registers all tools, resources, and prompts
- The MCP endpoint becomes available for client connections
Real-Time Log Viewer
The Logs tab on the Server page shows live output from the MCP server process. Logs are streamed via WebSocket. Use the log viewer to:- Verify the server started successfully
- Debug issues with specific tools or resources
- Monitor incoming MCP client requests
Server Status
| Status | Description |
|---|---|
| Running | MCP server is active and accepting connections |
| Stopped | MCP server is not running |
Stop and Restart
- Stop: Click Stop to shut down the running MCP server
- Restart: Click Deploy again to regenerate code and start a fresh instance
Connecting MCP Clients
After deployment, connect MCP clients using the endpoint URL displayed on the Server page.Claude Desktop
Add to yourclaude_desktop_config.json:
Cursor
Add to.cursor/mcp.json:
Claude Code
MCP Endpoint URL
<host> with your server’s hostname and <port> with the MCP port (default: 8080).
If SSL is enabled, use https://. See SSL/HTTPS Setup for details.
See Also
Tools
Build the SQL-backed tools that get deployed
Resources
Create the MCP resources included in deployment
SSL/HTTPS Setup
Enable HTTPS for the MCP endpoint
Queue Management
Control concurrent query execution per connection
Agent MCP
Give AI agents programmatic access to your Studio
Troubleshooting
Common deployment issues and solutions