What is the primary function of "sampling" when connecting an MCP server to a language model?	To shift the responsibility of generating text from the server to the client.
What is a major benefit of using sampling when building a publicly accessible MCP server?	The MCP server does not need an API key, and the cost of generated tokens is handled by the client.
What is the complexity drawback of having the MCP server directly access the LLM (Option 1)?	It requires writing code on the server to handle API calls, extract generated text, and manage API keys.
On the MCP server side, what function must be called within a tool function to initiate the sampling process?	The `create message` function, passing a list of messages.
What specific piece of code must be written on the client side to handle the messages sent from the server during sampling?	A sampling callback.
In the sampling process, which component is responsible for actually calling the LLM (like Claude) to generate the text?	The MCP client.
Conceptually, what does the technique of sampling represent?	Moving the burden of text generation from the server back over to the client.
When should a developer prioritize using the sampling technique?	Anytime they are building a publicly accessible MCP server.