inprocess_sampling

README

InProcess Sampling Example

This example demonstrates how to use sampling with in-process MCP client/server communication.

Overview

The example shows:

• Creating an MCP server with sampling enabled
• Adding a tool that uses sampling to request LLM completions
• Creating an in-process client with a sampling handler
• Making tool calls that trigger sampling requests

Key Components

Server Side

• mcpServer.EnableSampling() - Enables sampling capability
• Tool handler calls mcpServer.RequestSampling() to request LLM completions
• Sampling requests are handled directly by the client's sampling handler

Client Side

• MockSamplingHandler - Implements the SamplingHandler interface
• NewInProcessClientWithSamplingHandler() - Creates client with sampling support
• The handler receives sampling requests and returns mock LLM responses

Running the Example

go run main.go

Expected Output

Tool result: LLM Response (model: mock-llm-v1): Mock LLM response to: 'What is the capital of France?'

Real LLM Integration

To integrate with a real LLM service (OpenAI, Anthropic, etc.), replace the MockSamplingHandler with an implementation that calls your preferred LLM API. See the client sampling documentation for examples with real LLM providers.