
ALT While hashtag#MCP (Model Context Protocol) is promising for hashtag#AI, it still has critical gaps that need addressing before it can be of practical use in enterprise scenarios: 1. Currently only the desktop version of LLMs can leverage MCP, which means there is lot of catching up to do by the cloud versions. 2. The term "MCP Server" is misleading - it is not a standalone server as one might expect from its name but just a code module tightly coupled to client applications. This means the clients have to manually load and run the module, creating operational friction and security vulnerabilities. Not to mention there is no explicit standard on the client and MCP server interaction protocols. 3. MCP's schema definitions reinvent the wheel instead of aligning with existing standards (e.g., OpenAPI). Critical metadata like function return types are missing in the definition, which means the LLMs have to figure out how to marshal data between formats. https://gk.palem.in/articles/























