When does MCP make sense?

When does using MCP make sense? My latest for The New Stack cuts through the hype to answer this loaded question head-on.

The general consensus is that MCP is a "force multiplier" for indeterministic workflows, novel integrations, and giving AI coding agents context on the fly.

But for more predictable automation, typical CLI commands or API integrations are going to be more reliable in the long run. Anything else is probably overengineering.

Also, one key best practice: You shouldn't be mapping APIs one-to-one — MCP tools should model user intents and consider natural language.

My piece goes a bit deeper and looks at some emerging MCP server success stories at GitHub, Atlassian, Red Hat, and beyond.

This is definitely an evolving area, but I hope this piece tries to make sense of when it's best to put MCP to use!