What is the Model Context Protocol (MCP)?
An open standard for LLM apps to connect to external tools, data sources, and workflows
Originally open-sourced by Anthropic, it allows for consistent communication between clients and servers.
What does MCP function like in the context of AI apps?
USB-C for AI apps
It enables multiple integrations without the need for custom solutions each time.
What are the three core surfaces of MCP?
These surfaces define the types of interactions and data exchanges possible within the protocol.
What type of messages does MCP use for communication?
JSON-RPC 2.0
Common transports include stdio for local tools and streamable HTTP for remote ones.
What is one of the key benefits of MCP?
Solves the N×M integration problem
It allows many models to communicate with many tools without needing custom integrations.
What does MCP enable in terms of data fetching?
Richer, cheaper context
Clients can fetch exactly what’s needed from MCP servers, reducing token load.
Name a few real use cases for MCP.
These applications demonstrate the versatility of the protocol in various environments.
True or false: MCP includes built-in security features for authentication and hardening.
FALSE
Authentication and hardening are the responsibility of the user; teams should implement security measures like least-privilege credentials.
Where can you find the spec & reference for MCP?
modelcontextprotocol.io + GitHub
This includes schemas and examples for implementation.
What should you use to expose your own tools/resources/prompts in MCP?
stdio for local, streamable HTTP/SSE for remote
It’s important to secure these with API keys or OAuth.
What is an MCP server?
A program that implements the Model Context Protocol and exposes tools, resources, and prompts over a standard JSON-RPC interface
It allows MCP-capable clients to discover and call these surfaces.
What does an MCP server do?
It focuses on the protocol, not identity.
What are the common transport methods for an MCP server?
These methods are used for local or remote access.
What is the role of auth/permissions in an MCP server?
Left to your deployment, focusing on least privilege and approvals
MCP does not handle identity.
How do clients interact with an MCP server?
They connect to one or more servers and route model requests to them
This solves the N×M ‘every model to every tool’ problem.
Name some examples of official/community servers for MCP.
These servers demonstrate safe brokering of systems to an AI app.
What are the first two steps in building an MCP server?
These steps are crucial for setting up the server.
Fill in the blank: An MCP server handles calls and streams results back to the _______.
client
This is a key function of the MCP server.
What is the primary purpose of an MCP server?
Let AI clients call tools, read resources, and use prompts in a standardized way
MCP servers facilitate interaction between AI applications and various tools/resources.
What are the three distinct branches of the United States government?
Each branch has specific powers and responsibilities to check the powers of the others, ensuring a balance of power.
What is the interface specification used by MCP servers?
Model Context Protocol (JSON-RPC 2.0 messages + MCP semantics for tools/resources/prompts)
This specification defines how AI clients communicate with MCP servers.
What type of operation model does an MCP server use?
Procedure calls (“run this tool with these params”) + streaming results/tokens
This model allows for dynamic interaction with tools and resources.
What are the transports used by MCP servers?
These transports enable various methods of communication with the MCP server.