Most MCP servers just wrap CRUD JSON APIs into tools — I did it too with scim-mcp and garmin-mcp-app. It works, until you realize a tool call dumps 50KB+ into context.
MCP isn't dead — but we need to design MCP tools with the context window in mind.
Most MCP servers just wrap CRUD JSON APIs into tools — I did it too with scim-mcp and garmin-mcp-app. It works, until you realize a tool call dumps 50KB+ into context.
MCP isn't dead — but we need to design MCP tools with the context window in mind.
| #include <stdio.h> | |
| #include <stdlib.h> | |
| #include <stdint.h> | |
| #ifdef _MSC_VER | |
| #include <intrin.h> /* for rdtscp and clflush */ | |
| #pragma optimize("gt",on) | |
| #else | |
| #include <x86intrin.h> /* for rdtscp and clflush */ | |
| #endif |
| #include <type_traits> | |
| #include <cstdio> | |
| namespace L { | |
| template <int n> | |
| struct N { | |
| static constexpr bool equals(int m) { return m == n; } | |
| }; |
Dear C++ library writer,
If your library forces me to use new all over, I will hate you.
If your library has types with bogus values, I will hate you.
If the documentation for your library gets the terminology of its own domain wrong, I will hate you.
If I say "My God, it's full of stars!" when I see the function signatures in your library, I will hate you.