tracing

Tracing Example

This example demonstrates how to use the agent SDK with tracing capabilities using both Langfuse and OpenTelemetry.

Prerequisites

• Go 1.21 or later
• Langfuse account and API keys
• OpenTelemetry collector running locally (or a remote endpoint)

Environment Variables

Set the following environment variables before running the example:

# Langfuse configuration
export LANGFUSE_SECRET_KEY="your-langfuse-secret-key"
export LANGFUSE_PUBLIC_KEY="your-langfuse-public-key"
export ENVIRONMENT="development"

# OpenTelemetry configuration
export SERVICE_NAME="agent-sdk-example"
export OTEL_COLLECTOR_ENDPOINT="localhost:4317"  # Default OpenTelemetry collector endpoint

# OpenAI configuration
export OPENAI_API_KEY="your-openai-api-key"
export LLM_MODEL="gpt-4"  # or your preferred model

# System prompt
export SYSTEM_PROMPT="You are a helpful AI assistant."

Running the Example

1. Start the OpenTelemetry collector (if running locally):

docker run -v $(pwd)/otel-collector-config.yaml:/etc/otel-collector-config.yaml \
    -p 4317:4317 \
    -p 4318:4318 \
    otel/opentelemetry-collector:latest \
    --config /etc/otel-collector-config.yaml

1. Run the example:

go run cmd/examples/tracing/main.go

1. Interact with the agent by entering queries. Type 'exit' to quit.

Code Structure

The example consists of two main files:

• main.go: Contains the main program that sets up tracing and runs an interactive agent
• production_config.go: Contains production-ready configuration functions that can be reused in other applications

Using in Your Application

To use the tracing configuration in your own application:

import (
    "context"
    "github.com/Ingenimax/agent-sdk-go/cmd/examples/tracing"
)

func main() {
    ctx := context.Background()

    // Create a traced agent
    agent, err := tracing.CreateTracedAgent(ctx)
    if err != nil {
        log.Fatalf("Failed to create agent: %v", err)
    }

    // Use the agent
    response, err := agent.Run(ctx, "Your query here")
    if err != nil {
        log.Printf("Error: %v", err)
        return
    }

    fmt.Println(response)
}

Tracing Features

The example demonstrates:

1. Langfuse Integration

   • Tracks LLM calls and responses
   • Monitors performance metrics
   • Provides a web interface for visualization

2. OpenTelemetry Integration

   • Distributed tracing
   • Metrics collection
   • Integration with various observability platforms

3. Middleware Pattern

   • LLM tracing middleware
   • Memory tracing middleware
   • Extensible for additional tracing needs

Viewing Traces

1. Langfuse Dashboard

   • Visit https://cloud.langfuse.com
   • Navigate to your project
   • View traces, metrics, and logs

2. OpenTelemetry

   • Use tools like Jaeger, Zipkin, or your preferred observability platform
   • Connect to your OpenTelemetry collector endpoint
   • View distributed traces and metrics