This directory contains examples of how to use the Helicone Python SDK to log tool usage in your LLM applications.

1. Install the required package:

   pip install helicone-helpers

2. Set your Helicone API key as an environment variable (optional):

   export HELICONE_API_KEY="your-helicone-api-key"

The simple_example.py file contains a basic calculator tool example that demonstrates the core functionality of logging tool usage with Helicone.

python simple_example.py

The example.py file contains more complex examples including:

• Weather API tool (with mock data)
• Database query tool (simulated)
• Document search tool (simulated)

python example.py

1. HeliconeManualLogger: The main class for logging tool usage
2. Request Object: Must include _type: "tool", toolName, and input
3. Operation Function: The function that performs the tool operation and logs results
4. Result Recorder: Used to append results to be logged to Helicone

# Initialize the logger
helicone_logger = HeliconeManualLogger(api_key="your-api-key")

# Define the operation function
def tool_operation(result_recorder):
    # Access input from result_recorder.request
    # Perform the operation
    # Log results with result_recorder.append_results()
    return result

# Log the request
result = helicone_logger.log_request(
    request={
        "_type": "tool",
        "toolName": "your-tool-name",
        "input": { ... }
    },
    operation=tool_operation
)

• Session Management: Use Helicone-Property-Session-Id in additional headers
• User Tracking: Use Helicone-Property-User-Id in additional headers
• Custom Properties: Add any custom properties with Helicone-Property-* headers

For more information, see the Helicone Tools Python SDK documentation.