​
1. Installation

pip install mcp-logger

​
2. Initialize the SDK within your main MCP server file

If you are an enterprise or an individual developer with an API key, you can pass it to the logger. This will enable you to see all input and response data from tools and prompts. 
from observee import ObserveeConfig, observee_usage_logger, observee_prompt_logger

# Set the MCP server name globally
ObserveeConfig.set_mcp_server_name("your-mcp-server-name")

# Optionally, if you have an API key, you can pass it to the logger.
ObserveeConfig.set_api_key("your-api-key")

​
3. Add logging to tool calls.

In Python, you can add it to each individual tool call, or you can use the observee_usage_logger decorator to log all tool calls. 
@observee_usage_logger
async def your_tool_function():
    # Your tool code here
    pass

​
4. Add logging to your prompt calls.

@observee_prompt_logger
async def your_prompt_function():
    # Your prompt code here
    pass