Using your first MCP server

Learn how to connect your MCP server to an LLM provider and build a chat-enabled app

Now that you’ve deployed your first MCP server and confirmed it’s working, you can connect it to an LLM provider like OpenAI.

In this guide, you’ll learn how to build a chat-enabled app that automatically handles tool calls from your Metorial-powered MCP server.

What you’ll learn:

• How to use a Metorial MCP server
• How to use the Metorial SDKs

Before you start:

• Complete the Introduction guide
• Complete the Deploying your first MCP server guide
• Complete the Testing your first MCP server guide

1. 1. Install the SDKs

   Run the installer for your language of choice:

   TypeScript

   npm install metorial @metorial/openai openai
   
2. 2. Configure Clients

   Instantiate both clients with your API keys and your MCP server ID.

   TypeScript

   import Metorial from 'metorial';
   import OpenAI from 'openai';
   
   const metorial = new Metorial({
     apiKey: '$$SECRET_TOKEN$$'
   });
   const openai = new OpenAI({
     apiKey: '...your-openai-api-key...'
   });
   
3. 3. Fetch Your Server Tools

   Retrieve the session object that exposes your deployed MCP tools.

   TypeScript

   const session = await metorial.withProviderSession(
     metorialOpenAI.chatCompletions,
     { serverDeployments: ['...server-deployment-id...'] }
   );
   
4. 4. Send Your First Prompt

   Kick off the loop by sending an initial message.

   TypeScript

   let messages = [
     { role: "user", content: "Summarize the README.md file of the metorial/websocket-explorer repository on GitHub." }
   ];
   
5. 5. Loop & Handle Tool Calls

   1. Send messages to OpenAI, passing tools: session.tools (TS) or tools=session.tools (Py).
   2. If the assistant response contains tool_calls, invoke it:

   TypeScript

   const response = await openai.chat.completions.create({
     model: 'gpt-4o',
     messages,
     tools: session.tools
   });
   const choice = response.choices[0]!;
   const toolCalls = choice.message.tool_calls;
   const toolResults = await session.callTools(toolCalls);
   

   3. Append both the tool call requests and their results to messages.
   4. Repeat until the assistant’s response has no more tool_calls.
6. 6. Display the Final Output

   Once there are no more tool calls, your assistant’s final reply is in:

   TypeScript

   console.log(choice.message.content);
   

What’s Next?

You are all set on having a production-ready MCP server to use in your AI apps. Next, you will learn about all the dev tooling available.

Next Up: How to monitor your server and tool calls

Learn how to use the observability & logging features.