MCP Sampling

[Gerr1ck]

Question

Hello,

According to the Model Context Protocol (MCP) specification (Sampling), the MCP Server can forward prompts to the client for execution at a large language model (LLM) through a sampling mechanism.

I am trying to implement a use case where the MCP Server performs "sampling" and requests the MCP Client to execute a prompt and return the output without involving a User-In-The-Loop.

Essentially, I want the MCP Server to trigger prompt execution independently and receive the result from the client, and that the MCP Server can trigger prompt execution whenever the server needs it.

I have some questions to clarify this setup:

1. Is it correct that the MCP Server can independently initiate a sampling message to the MCP Client, without the MCP Client first invoking a TOOL (or something else) on the MCP Server?

2. If this is possible, could you please provide example code or guidance demonstrating how to implement this flow where the MCP Server triggers sampling and the client responds with the prompt output?

Thank you for your assistance.

Additional Context

No response

[maxisbey]

1. Yes the MCP server can initiate a sampling request whenever it wants without the client having first sent a tool call. Although, this does require a stateful connection to be open, whether that be the now deprecated SSE transport, or Streamable HTTP in stateful mode.
2. We're currently pretty lacking on docs for sampling, see here for a recent PR adding examples: [DOC] simple mcp client with sampling capability #1436

[Gerr1ck]

@maxisbey okay understood. I need to use a stateful connection like Streamable HTTP. Before, I was trying to implement this Use Case with STDIO...

Regarding to the second bulletpoint: This only covers a MCP Client which enables sampling capability, as far I understood. Thats only one piece.

The other interesting part would be how a MCP Server sends a sampling message to the client. How can I implement this? Any hints which methods from the SDK shall I use for this?

[yarnabrina]

Hi @Gerr1ck , I'm not familiar with any official docs but if a dummy sample is enough for you to get a first look (this is what I tested the PR client initially with), you can take a look here.

https://github.com/yarnabrina/learn-model-context-protocol/blob/main/src/mcp_learning/mcp_server/simplification.py#L13-L66

(This is not stateful though.)

[Gerr1ck]

Hi @yarnabrina thank you the link.

Question to that:
I cannot find a reference where the method
async def parse_arithmetic_expression(text: str, context: Context) -> str:
gets called.

Can you explain me how this method gets called and which context shall be used?

Imagine that the MCP Server just want to call parse_arithmetic_expression itself. From where do the MCP Server gets the context?

[yarnabrina]

There is no explicit call to parse_arithmetic_expression. It's an MCP tool, and converted and exposed that way here. I could have used .tool decorator as well, would have worked similarly.

It never needs to be called explicitly, based on the available tools and contracts, LLM decides what to call with what argument and then MCP client calls the same. You can use my repo above for client example, or the PR I created also have a minimal version.

I'm not exactly familiar with context creation, I believe it's set at low level, and it's part of the contract if you want to use sampling/elicitation. But hopefully @maxisbey or @felixweinberger can answer in details more, I'm not too familiar at the low level details.