4.5 Starting and Managing the SQLcl MCP Server

The SQLcl MCP Server is automatically launched and managed when you start SQL Developer.

After the MCP server is initialized, you can access the tools it provides through the GitHub Copilot Chat Extension in Agent mode. You can interact with the agent using either predefined commands or natural language input. Predefined commands help you perform specific actions quickly, while natural language input allows you to type your requests in plain English, which the agent will understand and respond to accordingly.

4.5.1 Accessing SQLcl MCP Server

The SQLcl MCP Server is automatically initialized when you start SQL Developer.

Perform the following steps to access the SQLcl MCP Server:
  1. Click the Copilot Chat button.
    The Edit with Copilot chat window appears in the right pane.
  2. From the chat mode drop-down list, select Agent.

4.5.2 Accessing SQLcl MCP Server Tools

The SQLcl MCP Server provides essential tools for performing basic SQL operations, including connecting to an Oracle database, executing SQL and PL/SQL statements, and retrieving and analyzing results.

Perform the following steps to access the MCP Server tools:
  1. Access the SQLcl MCP Server. See Accessing SQLcl MCP Server.
  2. In the Copilot chat view, click the tools button to list the available tools.
  3. Select or deselect the tools you want to use.
    For more information about SQLcl MCP Server tools, see About the SQLcl MCP Server Tools.
    You can also search for any tool by typing its name in the search box.

4.5.3 Using the Agent Mode

You can interact with the agent in VS Code by using predefined commands or natural language input. Follow these steps to get started:

  1. Open the Chat view. See Accessing SQLcl MCP Server.
  2. Enter a predefined command or type a natural language query and click the Send button.
    Copilot requests confirmation to continue using the tool.
  3. Use the Continue button’s dropdown options to automatically confirm the selected tool for the current session, the workspace, or all future sessions.

    Note:

    You can click the chevron next to the tool name to view its details and input parameters.

    To learn more about Agent mode, see Use Agent Mode in VS Code.
After the task is executed, a green check mark is shown next to the tool, along with the generated result: