Skip to main content

Observability

Langfuse integration for production observability. Trace every LLM call, track costs, debug issues, and monitor performance in real-time.

What Problem Does This Solve?​

Problem: Can't debug production AI issues Solution: Complete tracing and monitoring

Basic Usage​

import { LangfuseTracer } from 'clear-ai-v2/shared';

const tracer = new LangfuseTracer({
  publicKey: process.env.LANGFUSE_PUBLIC_KEY,
  secretKey: process.env.LANGFUSE_SECRET_KEY
});

// Start trace
const trace = tracer.startTrace('user_query', {
  userId: 'user_123',
  sessionId: 'session_456'
});

// Track LLM call
const generation = tracer.trackGeneration(trace.id, {
  name: 'chat',
  input: messages,
  model: 'gpt-3.5-turbo'
});

const response = await llm.chat(messages);

tracer.endGeneration(generation.id, {
  output: response,
  tokens: { input: 150, output: 75, total: 225 },
  cost: 0.0003
});

// End trace
tracer.endTrace(trace.id);

Dashboard​

View all traces at: https://cloud.langfuse.com

See:

  • Every prompt and response
  • Token usage and costs
  • Execution times
  • Error rates
  • User sessions

Testing​

Optional dependency - works with or without Langfuse.

Infrastructure complete! Next: Types