Menu

1 / 5

Configuring Tool Traces In Your MCP Gateway

DEV Community·Michael Levan·7 months ago
#VexD0DRC
#ai#programming#mcp#kubernetes#name
Reading 0:00
15s threshold

An Agent makes a call to an LLM. The LLM decides which MCP server tool should be used for a task. The Agent then makes a call to said tool. This can happen once, or it can happen hundreds of times.

Here's the question: Do you know what MCP Server tools were used, when they were used, and where the prompt originated from? In other words, how can you actually track and confirm tool traces within your MCP Gateway?

That's where having an MCP Gateway that exposes these traces and metrics comes into play.

In this blog, you'll learn how to do full end-to-end trace observability for any MCP Server and tool.

Prerequisites

To follow along with this blog post in a hands-on fashion, you will need:

A k8s cluster (local is fine).

Agentgateway OSS installed, which you can find here .

A GitHub account because you will need a PAT (personal access token) to use the GitHub Copilot MCP Server.

If you don't have a k8s cluster, there's a large chunk of this blog post that's pretty visual, so you can follow along from a…

Continue reading — create a free account

Join HashtagPLUS to read full articles, follow hashtags, vote, and join the conversation.