Single-Span Tracing - Lumina Documentation

Track individual LLM calls with automatic attribute extraction.

Basic Usage

const response = await lumina.traceLLM(
  () =>
    openai.chat.completions.create({
      model: 'gpt-4',
      messages: [{ role: 'user', content: 'Hello!' }],
    }),
  {
    name: 'chat-completion',
    system: 'openai',
    prompt: 'Hello!',
  }
);

Automatic Attributes

Lumina automatically extracts:

Model name
Token counts (prompt, completion)
Latency (milliseconds)
Cost (USD)
Status (success/error)
Response text

Custom Metadata

Add custom attributes:

await lumina.traceLLM(
  () => llm.generate(prompt),
  {
    name: 'chat',
    metadata: {
      userId: 'user-123',
      sessionId: 'session-456',
      feature: 'customer-support',
    },
  }
);

Next Steps

Multi-Span Tracing

Track complex workflows with hierarchical spans