Techtuesday Mcp Lab Guide

Last Tuesday, Anthropic donated MCP to the Linux Foundation.

If that sentence doesn't mean much to you yet, it might soon. The last time I saw a move like this was Docker donating their container runtime to the CNCF. And we know how that played out.

MCP (Model Context Protocol) is a standardized way for AI models to connect to external tools and data. Think of it as a universal adapter between LLMs and everything else like databases, APIs, your infrastructure, whatever.

Right now, if you want an AI assistant to query your systems, you're writing custom glue code for every integration. MCP is trying to solve the n×m problem the same way containers solved the "works on my machine" problem.

I spent some time this weekend building a simple MCP server that lets Claude query a local Kubernetes cluster. Ask it "what pods are failing?" and it checks for you. No copy-pasting kubectl output into a chat window.

The setup was simpler than I expected. A few decorators in Python, point Claude Desktop at it, done. The part that surprised me was how quickly it went from "toy example" to "wait, this could be useful."

I put together a step-by-step lab guide if you want to try it yourself. Takes a couple of hours, assumes basic Python and K8s familiarity. Link in comments.

What I'm still figuring out is where the line between "cool demo" and "production-ready tool" is when the standards haven't been established yet. The protocol is solid, but the ecosystem is early. Feels like Docker in 2014 where we're clearly going somewhere, but not entirely sure where yet.

Anyone else working with MCP? Curious what people are building with it.

#TechTuesday #LearnWithOJ #MCP #DevOps #SRE #Kubernetes

First comment: MCP Kubernetes server guide: https://github.com/oj-codes/techtuesday/tree/main/2025-12-16-mcp-kubernetes-server

Hi Anthony, I am open to contract roles starting in the new year. Let's chat.