> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mcp-use.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Sampling

> Enable LLM sampling capabilities for MCP tools

<Info>
  Sampling allows MCP tools to request LLM completions during their execution.
</Info>

## Configuration

To enable sampling, provide a `sampling_callback` function when initializing the MCPClient:

<CodeGroup>
  ```python Python theme={null}
  from mcp_use.client import MCPClient
  from mcp.client.session import ClientSession
  from mcp.types import (
      CreateMessageRequestParams,
      CreateMessageResult,
      ErrorData,
      TextContent
  )

  async def sampling_callback(
      context: ClientSession,
      params: CreateMessageRequestParams
  ) -> CreateMessageResult | ErrorData:
      """
      Your sampling callback implementation.
      This function receives a prompt and returns an LLM response.
      """
      # Integrate with your LLM of choice (OpenAI, Anthropic, etc.)
      response = await your_llm.complete(params.messages[-1].content.text)

      return CreateMessageResult(
          content=TextContent(text=response, type="text"),
          model="your-model-name",
          role="assistant"
      )

  # Initialize client with sampling support
  client = MCPClient(
      config="config.json",
      sampling_callback=sampling_callback
  )
  ```
</CodeGroup>

## Creating Sampling-Enabled Tools

When building MCP servers, tools can request sampling using the context parameter:

```python theme={null}
from fastmcp import Context, FastMCP

mcp = FastMCP(name="MyServer")

@mcp.tool
async def analyze_sentiment(text: str, ctx: Context) -> str:
    """Analyze the sentiment of text using the client's LLM."""
    prompt = f"""Analyze the sentiment of the following text as positive, negative, or neutral.
    Just output a single word - 'positive', 'negative', or 'neutral'.

    Text to analyze: {text}"""

    # Request LLM analysis through sampling
    response = await ctx.sample(prompt)

    return response.text.strip()
```

## Error Handling

If no sampling callback is provided but a tool requests sampling:

<CodeGroup>
  ```python Python theme={null}
  # Without sampling callback
  client = MCPClient(config="config.json")  # No sampling_callback

  # Tool that requires sampling will return an error
  result = await session.call_tool("analyze_sentiment", {"text": "Hello"})
  # result.isError will be True
  ```
</CodeGroup>