1. Set Up Your First Tracer

Problem: You need to integrate HoneyHive tracing into your LLM application quickly to start monitoring calls and performance.

Solution: Initialize a HoneyHive tracer with minimal configuration and verify it’s working in under 5 minutes.

This guide walks you through setting up your first tracer, making a traced LLM call, and verifying the trace appears in your HoneyHive dashboard.

1.1. Prerequisites

  • Python 3.11+ installed

  • HoneyHive API key (get one at https://app.honeyhive.ai)

  • A HoneyHive project created (or we’ll create one for you)

1.2. Installation

Install the HoneyHive SDK:

pip install honeyhive

For LLM provider integrations, install with the provider extra:

# For OpenAI
pip install honeyhive[openinference-openai]

# For Anthropic
pip install honeyhive[openinference-anthropic]

# For multiple providers
pip install honeyhive[openinference-openai,openinference-anthropic]

1.3. Step 1: Set Up Environment Variables

Create a .env file in your project root:

# HoneyHive configuration
HH_API_KEY=your-honeyhive-api-key
HH_PROJECT=my-first-project
HH_SOURCE=development

# Your LLM provider API key
OPENAI_API_KEY=your-openai-api-key

1.4. Step 2: Initialize Your First Tracer

Create a simple Python script to initialize the tracer:

from honeyhive import HoneyHiveTracer
from openinference.instrumentation.openai import OpenAIInstrumentor
import openai
import os

# Step 1: Initialize HoneyHive tracer (loads config from environment)
tracer = HoneyHiveTracer.init(
    project="my-first-project",  # Or use HH_PROJECT env var
    source="development"          # Or use HH_SOURCE env var
)  # API key loaded from HH_API_KEY

# Step 2: Initialize instrumentor with tracer provider
instrumentor = OpenAIInstrumentor()
instrumentor.instrument(tracer_provider=tracer.provider)

print("✅ Tracer initialized successfully!")

What’s happening here:

  1. HoneyHiveTracer.init() creates a tracer instance configured for your project

  2. OpenAIInstrumentor automatically captures OpenAI SDK calls

  3. instrumentor.instrument(tracer_provider=tracer.provider) connects the instrumentor to HoneyHive

1.5. Step 3: Make Your First Traced Call

Add a simple LLM call to test tracing:

# Make a traced OpenAI call
client = openai.OpenAI()  # Uses OPENAI_API_KEY from environment

response = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello! This is my first traced call."}
    ]
)

print(f"Response: {response.choices[0].message.content}")
print("✅ Trace sent to HoneyHive!")

Automatic tracing: Because the instrumentor is active, this call is automatically traced without any decorators or manual span creation.

1.6. Step 4: Verify in HoneyHive Dashboard

  1. Go to https://app.honeyhive.ai

  2. Navigate to your project (my-first-project)

  3. Click “Traces” in the left sidebar

  4. You should see your trace with: - Model: gpt-3.5-turbo - Input message: “Hello! This is my first traced call.” - Response from the model - Timing information - Token counts

Tip

Traces typically appear within 1-2 seconds. If you don’t see your trace:

  • Check that HH_API_KEY is set correctly

  • Verify your project name matches

  • Look for error messages in your Python output

1.7. Complete Example

Here’s the complete working script:

"""
first_tracer.py - Your first HoneyHive traced application

Run: python first_tracer.py
"""
from honeyhive import HoneyHiveTracer
from openinference.instrumentation.openai import OpenAIInstrumentor
import openai
import os
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

def main():
    # Initialize tracer
    tracer = HoneyHiveTracer.init(
        project="my-first-project",
        source="development"
    )

    # Initialize instrumentor
    instrumentor = OpenAIInstrumentor()
    instrumentor.instrument(tracer_provider=tracer.provider)

    print("✅ Tracer initialized!")

    # Make traced call
    client = openai.OpenAI()
    response = client.chat.completions.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": "Hello! This is my first traced call."}
        ]
    )

    print(f"\n📝 Response: {response.choices[0].message.content}")
    print("\n✅ Trace sent to HoneyHive!")
    print("👉 View at: https://app.honeyhive.ai")

if __name__ == "__main__":
    main()

1.8. Running the Example

# Install dependencies
pip install honeyhive[openinference-openai] python-dotenv

# Run the script
python first_tracer.py

Expected output:

✅ Tracer initialized!


📝 Response: Hello! I'm happy to help you with your first traced call...

✅ Trace sent to HoneyHive!
👉 View at: https://app.honeyhive.ai

1.9. Troubleshooting

Tracer initialization fails:

No traces appearing:

  • Wait 2-3 seconds for traces to process

  • Check project name matches in code and dashboard

  • Look for error messages in Python console

  • Verify instrumentor was initialized correctly

Import errors:

# Install the correct extras
pip install honeyhive[openinference-openai]

# Or install instrumentor directly
pip install honeyhive openinference-instrumentation-openai openai

1.10. Next Steps

Now that your tracer is working:

Completion time: ~5 minutes from installation to first trace ✨