CSE Roadmap

Core tier · about 5 hours · 14 parts

Part 10 of 14 · 1 min read

Answering MCP-architecture questions in an interview

How does MCP differ from plain function calling?

"Function calling is how the model asks for a call, in a provider-specific format, with tools that live inside your application. MCP is how an application finds and calls tools that live elsewhere, over a standard protocol. A host creates one client per server, the client discovers the server's tools, resources and prompts, and the host then shows the tools to the model as ordinary function schemas. They stack; MCP does not replace function calling. The trade is reuse and a clean boundary, against more moving parts and a larger attack surface, since the descriptions and results come from third parties." Follow-up: "What dates a weak answer here?" Not knowing that the protocol has an older handshake-and-session form and a newer stateless one, where every request carries its own version and capabilities. A client that only speaks one era cannot talk to every server.

How would you secure an MCP server that exposes internal data?

"Start from least privilege: narrow tools, scopes, row scope, caps, short-lived credentials. Those are the same controls that apply to any tool. Then add what MCP specifically brings: authorisation of the client, validating every input, treating server descriptions and annotations as untrusted in your own host, and logging every call." Follow-up: "Where does authorisation of the client fit into this picture?" It's the real token machinery behind the grant-shaped objects used here. It has a separate issuing service and an audience, so a token for one system cannot be replayed at another; module 5 builds it.