Skip to main content

LLM Trace Logging

Argus keeps trace artifacts for debugging OpenAI- and MiniMax-adjacent workflows without letting those traces define report truth.

Main uses​

  • correlate response IDs with case runs
  • inspect Stage 3A planning or synthesis output
  • compare deterministic report wording against model output

Artifact families​

  • artifacts/llm-traces/*
  • MiniMax comparison artifacts under artifacts/minimax-experiments/*