Replay Testing

Test prompt changes against real production traces before deployment.

Workflow

1. Capture baseline:

await lumina.createReplaySet({
  name: 'Pre-prompt-change',
  filters: {
    service: 'chat-api',
    timeRange: '24h',
    sampleSize: 100,
  },
});

2. Make changes: Update your prompts or code. 3. Replay:

await lumina.replayTraces({
  replaySetId: 'replay_001',
});

4. Compare: View side-by-side diffs with semantic scoring.

Replay makes real LLM API calls:

100 traces × $0.01 per call = $1.00

Use simulation mode for testing without costs:

await lumina.replayTraces({
  replaySetId: 'replay_001',
  mode: 'simulation',
});