get_current_local_time

LLM agents fail in production; tracing is key to fixing them

Production environments reveal LLM agent failures that synthetic evaluations often miss. Issues like infinite tool loops, runaway chains, and latency spikes from hidden retries are common in real-world deployments. Comp…