Integrate with Anthropic

Note

Problem-solving guide for Anthropic integration

This guide helps you solve specific problems when integrating HoneyHive with Anthropic, with support for multiple instrumentor options.

This guide covers Anthropic integration with HoneyHive’s BYOI architecture, supporting both OpenInference and Traceloop instrumentors.

Compatibility

Problem: I need to know if my Python version and Anthropic SDK version are compatible with HoneyHive.

Solution: Check the compatibility information below before installation.

Python Version Support

Support Level	Python Versions
Fully Supported	3.11, 3.12, 3.13
Not Supported	3.10 and below

Provider SDK Requirements

Minimum: anthropic >= 0.17.0
Recommended: anthropic >= 0.21.0
Tested Versions: 0.21.0, 0.22.0, 0.23.0

Instrumentor Compatibility

Instrumentor	Status	Notes
OpenInference	Fully Supported	Full Claude 3 family support with streaming and vision
Traceloop	Fully Supported	Enhanced metrics with Claude-specific cost tracking

Known Limitations

Streaming: Partial support - requires manual context management for proper traces
Vision API: Supported for Claude 3 models, traced automatically
Tool Use: Fully supported with both instrumentors
Message Batching: Not yet supported by instrumentors, use manual tracing

Note

For the complete compatibility matrix across all providers, see Multi-Provider Integration.

Choose Your Instrumentor

Problem: I need to choose between OpenInference and Traceloop for Anthropic integration.

Solution: Choose the instrumentor that best fits your needs:

OpenInference: Open-source, lightweight, great for getting started
Traceloop: Enhanced LLM metrics, cost tracking, production optimizations

Best for: Open-source projects, simple tracing needs, getting started quickly

# Recommended: Install with Anthropic integration
pip install honeyhive[openinference-anthropic]

# Alternative: Manual installation
pip install honeyhive openinference-instrumentation-anthropic anthropic>=0.17.0

from honeyhive import HoneyHiveTracer
from openinference.instrumentation.anthropic import AnthropicInstrumentor
import anthropic
import os

# Environment variables (recommended for production)
# .env file:
# HH_API_KEY=your-honeyhive-key
# ANTHROPIC_API_KEY=your-anthropic-key

# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)  # Uses HH_API_KEY from environment

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

# Basic usage with error handling
try:
    client = anthropic.Anthropic()  # Uses ANTHROPIC_API_KEY automatically
    response = client.messages.create(
        model="claude-3-sonnet-20240229",
        max_tokens=1000,
        messages=[{"role": "user", "content": "Hello!"}]
    )
    print(response.content[0].text)
    # Automatically traced! ✨
except anthropic.APIError as e:
    print(f"Anthropic API error: {e}")
except Exception as e:
    print(f"Unexpected error: {e}")

from honeyhive import HoneyHiveTracer, trace, enrich_span
from honeyhive.models import EventType
from openinference.instrumentation.anthropic import AnthropicInstrumentor
import anthropic

# Initialize with custom configuration
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    api_key="your-honeyhive-key",  # Or set HH_API_KEY environment variable
    project="your-project",        # Or set HH_PROJECT environment variable
    source="production"            # Or set HH_SOURCE environment variable
)

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

@trace(tracer=tracer, event_type=EventType.chain)
def analyze_document(document: str) -> dict:
    """Advanced example with business context and multiple Anthropic calls."""
    client = anthropic.Anthropic()

    # Add business context to the trace
    enrich_span({
        "business.input_type": type(document).__name__,
        "business.use_case": "document_analysis",
        "anthropic.strategy": "claude_reasoning",
        "instrumentor.type": "openinference"
    })

    try:
        # First call: Quick summary with Claude Sonnet
        summary_response = client.messages.create(
            model="claude-3-sonnet-20240229",
            max_tokens=500,
            messages=[{
                "role": "user",
                "content": f"Provide a brief summary of this document: {document}"
            }]
        )

        # Second call: Detailed analysis with Claude Opus
        analysis_response = client.messages.create(
            model="claude-3-opus-20240229",
            max_tokens=1000,
            messages=[{
                "role": "user",
                "content": f"Provide detailed analysis with insights: {document}"
            }]
        )

        # Add result metadata
        enrich_span({
            "business.successful": True,
            "anthropic.models_used": ["claude-3-sonnet-20240229", "claude-3-opus-20240229"],
            "business.result_confidence": "high"
        })

        return {"summary": summary_response.content[0].text, "analysis": analysis_response.content[0].text}

    except anthropic.APIError as e:
        enrich_span({
            "error.type": "api_error",
            "error.message": str(e),
            "instrumentor.source": "openinference"
        })
        raise

Common OpenInference Issues:

Missing Traces

# Use correct initialization pattern
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

Performance for High Volume

# OpenInference uses efficient span processors automatically
# No additional configuration needed

Multiple Instrumentors

# You can combine OpenInference with other instrumentors
from openinference.instrumentation.anthropic import AnthropicInstrumentor
 from openinference.instrumentation.openai import OpenAIInstrumentor

 # Step 1: Initialize HoneyHive tracer first (without instrumentors)
 tracer = HoneyHiveTracer.init(
     project="your-project"  # Or set HH_PROJECT environment variable
 )

 # Step 2: Initialize instrumentors separately with tracer_provider
 anthropic_instrumentor = AnthropicInstrumentor()
 openai_instrumentor = OpenAIInstrumentor()

anthropic_instrumentor.instrument(tracer_provider=tracer.provider)
openai_instrumentor.instrument(tracer_provider=tracer.provider)

Environment Configuration

# HoneyHive configuration
export HH_API_KEY="your-honeyhive-api-key"
export HH_SOURCE="production"

# Anthropic configuration
export ANTHROPIC_API_KEY="your-anthropic-api-key"

Best for: Production deployments, cost tracking, enhanced LLM observability

# Recommended: Install with Traceloop Anthropic integration
pip install honeyhive[traceloop-anthropic]

# Alternative: Manual installation
pip install honeyhive opentelemetry-instrumentation-anthropic anthropic>=0.17.0

from honeyhive import HoneyHiveTracer
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
import anthropic
import os

# Environment variables (recommended for production)
# .env file:
# HH_API_KEY=your-honeyhive-key
# ANTHROPIC_API_KEY=your-anthropic-key

# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)  # Uses HH_API_KEY from environment

# Step 2: Initialize Traceloop instrumentor separately with tracer_provider
instrumentor = AnthropicInstrumentor()
instrumentor.instrument(tracer_provider=tracer.provider)

# Basic usage with automatic tracing
try:
    client = anthropic.Anthropic()  # Uses ANTHROPIC_API_KEY automatically
    response = client.messages.create(
        model="claude-3-sonnet-20240229",
        max_tokens=1000,
        messages=[{"role": "user", "content": "Hello!"}]
    )
    print(response.content[0].text)
    # Automatically traced by Traceloop with enhanced metrics! ✨
except anthropic.APIError as e:
    print(f"Anthropic API error: {e}")
except Exception as e:
    print(f"Unexpected error: {e}")

from honeyhive import HoneyHiveTracer, trace, enrich_span
from honeyhive.models import EventType
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
import anthropic

# Initialize HoneyHive with Traceloop instrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    api_key="your-honeyhive-key",  # Or set HH_API_KEY environment variable
    project="your-project",        # Or set HH_PROJECT environment variable
    source="production"            # Or set HH_SOURCE environment variable
)

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

@trace(tracer=tracer, event_type=EventType.chain)
def analyze_document(document: str) -> dict:
    """Advanced example with business context and enhanced LLM metrics."""
    client = anthropic.Anthropic()

    # Add business context to the trace
    enrich_span({
        "business.input_type": type(document).__name__,
        "business.use_case": "document_analysis",
        "anthropic.strategy": "cost_optimized_claude_reasoning",
        "instrumentor.type": "openllmetry",
        "observability.enhanced": True
    })

    try:
        # First call: Quick summary with Claude Sonnet
        summary_response = client.messages.create(
            model="claude-3-sonnet-20240229",
            max_tokens=500,
            messages=[{
                "role": "user",
                "content": f"Provide a brief summary of this document: {document}"
            }]
        )

        # Second call: Detailed analysis with Claude Opus
        analysis_response = client.messages.create(
            model="claude-3-opus-20240229",
            max_tokens=1000,
            messages=[{
                "role": "user",
                "content": f"Provide detailed analysis with insights: {document}"
            }]
        )

        # Add result metadata
        enrich_span({
            "business.successful": True,
            "anthropic.models_used": ["claude-3-sonnet-20240229", "claude-3-opus-20240229"],
            "business.result_confidence": "high",
            "openllmetry.cost_tracking": "enabled",
            "openllmetry.token_metrics": "captured"
        })

        return {"summary": summary_response.content[0].text, "analysis": analysis_response.content[0].text}

    except anthropic.APIError as e:
        enrich_span({
            "error.type": "api_error",
            "error.message": str(e),
            "instrumentor.error_handling": "openllmetry"
        })
        raise

Common Traceloop Issues:

Missing Traces

# Ensure Traceloop instrumentor is passed to tracer
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor

# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

Enhanced Metrics Not Showing

# Ensure you're using the latest version
# pip install --upgrade opentelemetry-instrumentation-anthropic

# The instrumentor automatically captures enhanced metrics
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

Multiple Traceloop Instrumentors

# You can combine multiple Traceloop instrumentors
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
 from opentelemetry.instrumentation.openai import OpenAIInstrumentor

 # Step 1: Initialize HoneyHive tracer first (without instrumentors)
 tracer = HoneyHiveTracer.init(
     project="your-project"  # Or set HH_PROJECT environment variable
 )

 # Step 2: Initialize instrumentors separately with tracer_provider
 anthropic_instrumentor = AnthropicInstrumentor()      # Traceloop Anthropic
 openai_instrumentor = OpenAIInstrumentor()          # Traceloop OpenAI

 anthropic_instrumentor.instrument(tracer_provider=tracer.provider)
 openai_instrumentor.instrument(tracer_provider=tracer.provider)

Performance Optimization

# Traceloop instrumentors handle batching automatically
# No additional configuration needed for performance

Environment Configuration

# HoneyHive configuration
export HH_API_KEY="your-honeyhive-api-key"
export HH_SOURCE="production"

# Anthropic configuration
export ANTHROPIC_API_KEY="your-anthropic-api-key"

# Optional: Traceloop cloud features
export TRACELOOP_API_KEY="your-traceloop-key"
export TRACELOOP_BASE_URL="https://api.traceloop.com"

Comparison: OpenInference vs Traceloop for Anthropic

Feature Comparison
Feature	OpenInference	Traceloop
Setup Complexity	Simple, single instrumentor	Single instrumentor setup
Token Tracking	Basic span attributes	Detailed token metrics + costs
Model Metrics	Model name, basic timing	Cost per model, latency analysis
Performance	Lightweight, fast	Optimized with smart batching
Cost Analysis	Manual calculation needed	Automatic cost per request
Production Ready	✅ Yes	✅ Yes, with cost insights
Debugging	Standard OpenTelemetry	Enhanced LLM-specific debug
Best For	Simple integrations, dev	Production, cost optimization

Migration Between Instrumentors

From OpenInference to Traceloop:

# Before (OpenInference)
from openinference.instrumentation.anthropic import AnthropicInstrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

# After (Traceloop) - different instrumentor package
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

From Traceloop to OpenInference:

# Before (Traceloop)
from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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

# After (OpenInference)
from openinference.instrumentation.anthropic import AnthropicInstrumentor
# Step 1: Initialize HoneyHive tracer first (without instrumentors)
tracer = HoneyHiveTracer.init(
    project="your-project"  # Or set HH_PROJECT environment variable
)

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