[do_anh_tu]

I think MCP is awesome, mainly because it forces devs to design the simplest possible tools/APIs/functions so even an average-performance LLM can use them correctly to get things done.

As developers, we often want everything to be rich, verbose, and customizable — but the reality is that for most users (and now for AIs acting on their behalf), simplicity wins every time. It’s like designing a great UI: the fewer ways you can get lost, the more people (or models) can actually use it productively.

If MCP ends up nudging the ecosystem toward small, well-defined, composable capabilities, that’s a win far beyond just “AI integration.”

[brookst]

I don’t like MCP because it relies on good faith from the plugin provider. It works great in closed, trusted environments but it cannot scale across trust boundaries.

It just begs for spam and fraud, with badly-behaving services advertising lowest-cost, highest-quality, totally amazing services. It feels like the web circa 1995… lots of implicit trust that isn’t sustainable.

[nlawalker]

Totally agree - the true source of all of the value here is the new incentive to write very simple services with very simple documentation and to make that documentation easily discoverable.

It fills a gap that exists in most service documentation: an easily discoverable page for developers (specifically, those who already know how to use their ecosystem of choice's HTTP APIs) that has a very short list of the service's most fundamental functionality with a simplified specification so they can go and play around with it.

[croes]

Too bad that the S in MCP stands for security