noogai03 a day ago

I find the idea of highly optimised MCP kind of funny. The protocol is so far from finished - the update for stateless MCP (which is our only hope at properly scaling it) isn't even released yet in the core protocol.

And even then - the LLMs are going to be so much slower than the MCP server, I've never quite understood the point of trying to optimise it. Even FastMCP is pretty good because everything is async.

But even given all that - this is very impressive :)

  • anuramat 3 hours ago

    > so much slower

    unless you're working on 10 features in parallel with smaller models in best-of-4 groups

  • orochimaaru a day ago

    The agents call the MCPs, which could reside on resource constrained environments or devices. Other situations like MCPs that run on your laptop are better off as self contained binaries (my personal and probably unpopular opinion ;) ).

    So MCP optimization may be something that’s targeted to those areas.

cyber_kinetist 12 hours ago

If they really cared about performance (which I think is the case since they're using C++ for this) - they should use something else than nholmann-json. It's simple to use, but has one of the most atrocious performance for a C++ JSON library. (The safest choice would probably be RapidJSON)