What MCP actually standardises
Tools, resources and prompts over a defined transport, so a capability written once works in every client that speaks it.
The Model Context Protocol is a client-server convention for giving a model access to things. Servers expose tools that can be called, resources that can be read, and prompts that can be inserted, over a transport that is either a local process on standard input and output or an HTTP endpoint. Clients, Claude, Cursor, editors, your own application, discover what a server offers and present it to the model.
The value is the N-times-M problem disappearing. Before, every tool needed an integration for every assistant; now a server you write once works anywhere the protocol is spoken. The same shift as any other protocol standardisation, at the point in the cycle where writing a server is still a small, comprehensible piece of work.
Worth being precise about the boundary: MCP does not make the model safe, authorised or correct. It defines how capabilities are described and invoked. Everything about who may call what, and what happens when they do, remains ordinary server engineering, which is exactly the framing that keeps you out of trouble.
You should now be able to
- Distinguish tools, resources and prompts
- Explain what the protocol removes from your integration work
Loading…