The way sampleLLM in server-everything assembles results is incompatible with the defined client response format

[aiqlcom]

Describe the bug

servers/src/everything/everything.ts

Lines 400 to 406 in 926d86c

	const result = await requestSampling(
	prompt,
	ToolName.SAMPLE_LLM,
	maxTokens,
	);
	return {
	content: [{ type: "text", text: `LLM sampling result: ${result}` }],

In the sampleLLM function within @modelcontextprotocol/server-everything, the result returned by the client is directly inserted into the text.

However, according to the official definition, this result should be an object, such as:

{
  model: string,  // Name of the model used
  stopReason?: "endTurn" | "stopSequence" | "maxTokens" | string,
  role: "user" | "assistant",
  content: {
    type: "text" | "image",
    text?: string,
    data?: string,
    mimeType?: string
  }
}

As a consequence, the final content returned will display as [object Object] instead of the actual content of the object.

To Reproduce

Steps to reproduce the behavior:

1. Initialize a Client
2. Defining a request handler for the client such as:

client.setRequestHandler(CreateMessageRequestSchema, async (request) => {
    console.log(request)
    return {
        model: "test-sampling-model",
        stopReason: "endTurn",
        role: "assistant",
        content: {
            type: "text",
            text: "This is the test message from client which use as sampling LLM",
        }
    };
});

3. Trigger call tool of sampleLLM

Expected behavior

After the server retrieves the result, it should extract and insert the necessary content, rather than embedding the entire result object into the string.

Logs
Console log:

{
    "content": [
        {
            "type": "text",
            "text": "LLM sampling result: [object Object]"
        }
    ]
}

Additional context
Examples of feasible fix proposals:
Proposal 1

            const result = await requestSampling(prompt, ToolName.SAMPLE_LLM, maxTokens);
            return {
                content: [{ type: "text", text: `LLM sampling result: ${result.content.text}` }],
            };

Proposal 2

            const result = await requestSampling(prompt, ToolName.SAMPLE_LLM, maxTokens);
            return {
                content: [result.content],
            };

If you confirm that this is an issue and agree with the proposed fix, I can submit a PR.

Alternatively, feel free to share any other suggestions or perspectives you might have.