cmdrdata-anthropic

The standard for AI customer intelligence - track every Claude call by customer, feature, or any dimension

Join hundreds of companies making customer-level AI tracking the default. One line of code to add complete visibility into your AI operations. Free during beta.

📊 Complete AI Intelligence Layer

cmdrdata-anthropic is the missing analytics layer for your AI-powered application:

Track Everything That Matters

• Customer Intelligence - Know exactly which customers use what features
• Metadata Everything - Tag usage by feature, experiment, team, region, or any dimension
• Usage Patterns - Understand how your AI is actually being used
• Real-time Analytics - Instant visibility into your AI operations

Built for Modern AI Apps

• One-line integration - Drop-in replacement for Anthropic SDK
• Zero latency overhead - Async tracking never blocks your API calls
• Unlimited custom fields - Track any metadata that matters to your business
• Privacy first - Your data never touches our servers (optional self-hosting)

What You Can Track

• Token usage by customer, feature, experiment, or any dimension
• Model usage patterns (Claude 3.5 Sonnet, Claude 3 Haiku, etc.)
• Customer behavior - Who uses what, when, and how much
• Custom metadata - Unlimited fields for your specific needs
• Performance metrics - Latency, errors, success rates by segment

🛡️ Production Ready

Extremely robust and reliable - Built for production environments with:

• Resilient Tracking: Claude calls succeed even if tracking fails.
• Non-blocking I/O: Fire-and-forget tracking never slows down your application.
• Automatic Retries: Failed tracking attempts are automatically retried with exponential backoff.
• Thread-Safe Context: Safely track usage across multi-threaded and async applications.
• Enterprise Security: API key sanitization and input validation.

🚀 Quick Start

Installation

pip install cmdrdata-anthropic

Basic Usage

# Before
import anthropic
client = anthropic.Anthropic(api_key="your-anthropic-key")

# After - same API, automatic tracking!
import cmdrdata_anthropic
client = cmdrdata_anthropic.TrackedAnthropic(
    api_key="your-anthropic-key",
    cmdrdata_api_key="your-cmdrdata-key"
)

# Same API as regular Anthropic client
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Hello, Claude!"}]
)

print(response.content)
# Usage automatically tracked to cmdrdata backend!

Async Support

import cmdrdata_anthropic

async def main():
    client = cmdrdata_anthropic.AsyncTrackedAnthropic(
        api_key="your-anthropic-key",
        cmdrdata_api_key="your-cmdrdata-key"
    )

    response = await client.messages.create(
        model="claude-3-5-sonnet-20241022",
        max_tokens=1000,
        messages=[{"role": "user", "content": "Hello!"}]
    )

    print(response.content)
    # Async usage tracking included!

🎯 Customer Context Management

Automatic Customer Tracking

from cmdrdata_anthropic.context import customer_context

# Set customer context for automatic tracking
with customer_context("customer-123"):
    response = client.messages.create(
        model="claude-3-5-sonnet-20241022",
        max_tokens=1000,
        messages=[{"role": "user", "content": "Help me code"}]
    )
    # Automatically tracked for customer-123!

# Or pass customer_id directly
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Hello"}],
    customer_id="customer-456"  # Direct customer ID
)

Manual Context Management

from cmdrdata_anthropic.context import set_customer_context, clear_customer_context

# Set context for current thread
set_customer_context("customer-789")

response = client.messages.create(...)  # Tracked for customer-789

# Clear context
clear_customer_context()

💎 Advanced Analytics with Custom Metadata

Track arbitrary metadata with each API call to enable sophisticated analytics:

# Example: AI writing assistant with feature tracking
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=2000,
    messages=[{"role": "user", "content": "Write a blog post about AI..."}],
    customer_id="customer-123",
    # Custom metadata for analytics
    custom_metadata={
        "feature": "content_generation",
        "experiment_group": "claude_3_5_test",
        "content_type": "blog_post",
        "user_segment": "power_user",
        "session_id": "sess_abc123"
    }
)

# Example: Customer support automation with behavior tracking
response = client.messages.create(
    model="claude-3-haiku-20240307",
    max_tokens=1000,
    messages=complex_support_conversation,
    customer_id="customer-456",
    custom_metadata={
        "use_case": "customer_support",
        "conversation_length": len(complex_support_conversation),
        "department": "technical_support",
        "interaction_type": "chat",
        "user_journey_stage": "resolution"
    }
)

Intelligence Use Cases:

• Feature adoption: Track which AI features customers actually use
• A/B testing: Compare model performance across experiment groups
• User segmentation: Understand usage patterns by customer segment
• Journey analytics: Track AI interactions throughout the customer journey
• Performance optimization: Identify which use cases need optimization
• Product insights: Data-driven decisions on feature development

⚙️ Configuration

Environment Variables

# Optional: Set via environment variables
export ANTHROPIC_API_KEY="your-anthropic-key"
export CMDRDATA_API_KEY="your-cmdrdata-key"
export CMDRDATA_ENDPOINT="https://api.cmdrdata.ai/api/events"  # Optional

# Then use without passing keys
client = cmdrdata_anthropic.TrackedAnthropic()

Custom Configuration

client = cmdrdata_anthropic.TrackedAnthropic(
    api_key="your-anthropic-key",
    cmdrdata_api_key="your-cmdrdata-key",
    cmdrdata_endpoint="https://your-custom-endpoint.com/api/events",
    track_usage=True,  # Enable/disable tracking
    timeout=30,  # Custom timeout
    max_retries=3  # Custom retry logic
)

🔒 Security & Privacy

Automatic Data Sanitization

• API keys automatically redacted from logs
• Sensitive data sanitized before transmission
• Input validation prevents injection attacks
• Secure defaults for all configuration

What Gets Tracked

# Tracked data (anonymized):
{
    "customer_id": "customer-123",
    "model": "claude-sonnet-4-20250514",
    "input_tokens": 25,
    "output_tokens": 150,
    "total_tokens": 175,
    "provider": "anthropic",
    "timestamp": "2025-01-15T10:30:00Z",
    "metadata": {
        "response_id": "msg_abc123",
        "type": "message",
        "stop_reason": "end_turn"
    }
}

Note: Message content is never tracked - only metadata and token counts.

📊 Monitoring & Performance

Built-in Performance Monitoring

# Get performance statistics
stats = client.get_performance_stats()
print(f"Average response time: {stats['api_calls']['avg']}ms")
print(f"Total API calls: {stats['api_calls']['count']}")

Health Monitoring

# Check tracking system health
tracker = client.get_usage_tracker()
health = tracker.get_health_status()
print(f"Tracking healthy: {health['healthy']}")

🛠️ Advanced Usage

Disable Tracking for Specific Calls

# Disable tracking for sensitive operations
response = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Private query"}],
    track_usage=False  # This call won't be tracked
)

Error Handling

from cmdrdata_anthropic.exceptions import CMDRDataError, TrackingError

try:
    client = cmdrdata_anthropic.TrackedAnthropic(
        api_key="invalid-key",
        cmdrdata_api_key="invalid-cmdrdata-key"
    )
except CMDRDataError as e:
    print(f"Configuration error: {e}")
    # Handle configuration issues

Integration with Existing Error Handling

# All original Anthropic exceptions work the same way
try:
    response = client.messages.create(...)
except anthropic.APIError as e:
    print(f"Anthropic API error: {e}")
    # Your existing error handling works unchanged

🔧 Development

Requirements

• Python 3.9+
• anthropic>=0.21.0

Installation for Development

git clone https://github.com/cmdrdata-ai/cmdrdata-anthropic.git
cd cmdrdata-anthropic
pip install -e .[dev]

Running Tests

# Run all tests
pytest

# Run with coverage
pytest --cov=cmdrdata_anthropic

# Run specific test categories
pytest -m unit          # Unit tests only
pytest -m integration   # Integration tests only

Code Quality

# Format code
black cmdrdata_anthropic/
isort cmdrdata_anthropic/

# Type checking
mypy cmdrdata_anthropic/

# Security scanning
safety check

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Workflow

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Add tests for your changes
5. Ensure all tests pass (pytest)
6. Format your code (black . && isort .)
7. Commit your changes (git commit -m 'Add amazing feature')
8. Push to the branch (git push origin feature/amazing-feature)
9. Open a Pull Request

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

🆘 Support

• Documentation: https://docs.cmdrdata.ai/anthropic
• Issues: GitHub Issues
• Support: spot@cmdrdata.ai

🔗 Related Projects

• cmdrdata-openai - Usage tracking for OpenAI
• CMDR Data Platform - Complete LLM usage analytics

📈 Changelog

See CHANGELOG.md for a complete list of changes and version history.

---

Built with ❤️ by the CMDR Data team

Become the Google Analytics of your AI - understand everything, optimize everything.