Why is Agent Skills Going to Kill MCP?

In this episode of the Naftiko Capabilities podcast, I sit down with Kevin Swiber to tackle one of the hottest debates in AI integration: are agent skills going to kill MCP? Spoiler — it’s more nuanced than that. Kevin breaks down the difference between MCP tools (the what) and agent skills (the how), explains why most enterprise teams are still using AI as a glorified Stack Overflow, and makes the case that skills are more accessible to non-developers than MCP servers will ever be. We also dig into the context window problem plaguing MCP hosts, why progressive disclosure matters, and when MCP is overkill for headless agents. If you’re trying to cut through the noise and figure out what actually matters for AI-powered workflows, this one’s for you.